Artificial Intelligence in The Gulf Challenges and Opportunities Elie Azar Full Chapter PDF

Artificial Intelligence in the Gulf:
Challenges and Opportunities Elie Azar

Visit to download the full and correct content document:
https://ebookmass.com/product/artificial-intelligence-in-the-gulf-challenges-and-opport
unities-elie-azar/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
The Ethics of Artificial Intelligence_ Principles,

Challenges, and Opportunities Prof Luciano Floridi
https://ebookmass.com/product/the-ethics-of-artificial-
intelligence_-principles-challenges-and-opportunities-prof-
luciano-floridi/
Oil and Oilseed Processing: Opportunities and

Challenges Tomás Lafarga
https://ebookmass.com/product/oil-and-oilseed-processing-
opportunities-and-challenges-tomas-lafarga/
International Management: Strategic Opportunities and

Cultural Challenges
https://ebookmass.com/product/international-management-strategic-
opportunities-and-cultural-challenges/
Artificial Intelligence and Data Science in

Environmental Sensing Mohsen Asadnia
https://ebookmass.com/product/artificial-intelligence-and-data-
science-in-environmental-sensing-mohsen-asadnia/
Artificial Intelligence, Game Theory and Mechanism
Design in International Politics Tshilidzi Marwala
https://ebookmass.com/product/artificial-intelligence-game-
theory-and-mechanism-design-in-international-politics-tshilidzi-
marwala/
Corporate Sustainability in Africa: Responsible

Leadership, Opportunities, and Challenges Samuel
Adomako
https://ebookmass.com/product/corporate-sustainability-in-africa-
responsible-leadership-opportunities-and-challenges-samuel-
adomako/
Artificial Intelligence and Intellectual Property Reto

Hilty
https://ebookmass.com/product/artificial-intelligence-and-
intellectual-property-reto-hilty/
Precision Health and Artificial Intelligence Arjun

Panesar
https://ebookmass.com/product/precision-health-and-artificial-
intelligence-arjun-panesar/
Blockchain in the Global South: Opportunities and

Challenges for Businesses and Societies Nir Kshetri
https://ebookmass.com/product/blockchain-in-the-global-south-
opportunities-and-challenges-for-businesses-and-societies-nir-
kshetri/
Artificial Intelligence
in the Gulf
Challenges and
Opportunities
Edited by
Elie Azar · Anthony N. Haddad
Artificial Intelligence in the Gulf
Elie Azar · Anthony N. Haddad
Editors
Artificial Intelligence
in the Gulf
Challenges and Opportunities
Editors
Elie Azar Anthony N. Haddad
Khalifa University of Science and Amazon
Technology Dubai, United Arab Emirates
Abu Dhabi, United Arab Emirates
ISBN 978-981-16-0770-7 ISBN 978-981-16-0771-4 (eBook)

https://doi.org/10.1007/978-981-16-0771-4
© Gulf Research Centre Cambridge 2021

This work is subject to copyright. All rights are solely and exclusively licensed by the
Publisher, whether the whole or part of the material is concerned, specifically the rights
of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on
microfilms or in any other physical way, and transmission or information storage and
retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc.
in this publication does not imply, even in the absence of a specific statement, that such
names are exempt from the relevant protective laws and regulations and therefore free for
general use.
The publisher, the authors and the editors are safe to assume that the advice and informa-
tion in this book are believed to be true and accurate at the date of publication. Neither
the publisher nor the authors or the editors give a warranty, expressed or implied, with
respect to the material contained herein or for any errors or omissions that may have been
made. The publisher remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.
Cover illustration: @Buena Vista Images/GettyImages
This Palgrave Macmillan imprint is published by the registered company Springer Nature
Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore
189721, Singapore
Acknowledgments
This volume is based on a selection of papers presented at the “Artificial

Intelligence in the Gulf: Prospects and Challenges” workshop, which was
held on July 15–18, 2019 at the University of Cambridge in the United
Kingdom. The co-editors would like to extend a warm thank you to all
participants of the workshop, who convened over three days to present
academic papers and discuss the prospects of Artificial Intelligence in the
six Gulf Cooperation Council member states. The discussions held during
the event greatly enriched the chapters of this volume and helped the
co-editors and authors together define a number of overarching themes,
challenges, and solutions relating to the topic.
The co-editors are also deeply grateful to the organizers of the 2019
Gulf Research Meeting of the Gulf Research Centre Cambridge, under
the auspices of which the workshop took place. In particular, we would
like to thank Dr. Abdulaziz Sager, Chairman and Founder of the Gulf
Research Centre (GRC), Dr. Oskar Ziemelis, Director of Cooperation at
the GRC, Dr. Christian Kokh, Director of the GRC Foundation Geneva,
as well as Ms. Aileen Byrne and Ms. Sanya Kapasi at the GRC, for their
tireless work and support.
United Arab Emirates Elie Azar

June 2021 Anthony N. Haddad
v
Contents
Part I Introduction
1 An Introduction to AI in the GCC 3
Elie Azar and Anthony N. Haddad
2 Framework of Study and Book Organization 9
Part II Data, Governance and Regulations

3 Public Sector Data for Academic Research: The Case
of the UAE 15
Aleksandar Abu Samra, Toufic Mezher, and Elie Azar
4 Strategy for Artificial Intelligence in Bahrain:
Challenges and Opportunities 47
Hesham Al-Ammal and Maan Aljawder
5 Thoughts and Reflections on the Case of Qatar:
Should Artificial Intelligence Be Regulated? 69
Ahmed Badran
vii
viii CONTENTS
Part III Existing Opportunities and Sectoral

Applications
6 Knowledge, Attitude, and Perceptions of Financial
Industry Employees Towards AI in the GCC Region 95
Muhammad Ashfaq and Usman Ayub
7 The GCC and Global Health Diplomacy: The New
Drive Towards Artificial Intelligence 117
Mohammed Sharfi
8 Free Zones in Dubai: Accelerators for Artificial
Intelligence in the Gulf 141
Robert Mogielnicki
Part IV Society, Utopia and Dystopia

9 AI & Well-Being: Can AI Make You Happy in the City 163
Ali al-Azzawi
10 Women and the Fourth Industrial Revolution:
An Examination of the UAE’s National AI Strategy 203
Victoria Heath
11 The Art and Science of User Exploitation: AI
in the UAE and Beyond 247
Helen Abadzi and Sahar ElAsad
12 Fatwas from Islamweb.Net on Robotics and Artificial
Intelligence 279
Julia Singer
Part V Conclusion
13 Outlook for the Future of AI in the GCC 305
Index 311
List of Contributors
Helen Abadzi University of Texas, Arlington, TX, USA

Aleksandar Abu Samra Khalifa University of Science and Technology,
Abu Dhabi, UAE
Hesham Al-Ammal University of Bahrain, Zallaq, Bahrain
Ali al-Azzawi Smart Dubai, Dubai, UAE
Maan Aljawder University of Bahrain, Zallaq, Bahrain
Muhammad Ashfaq IU International University of Applied Sciences,
Bad Honnef, Germany
Usman Ayub COMSATS University Islamabad, Islamabad, Pakistan
Elie Azar Khalifa University of Science and Technology, Abu Dhabi,
UAE
Ahmed Badran Department of International Affairs, Qatar University,
Doha, Qatar
Sahar ElAsad Regional Center for Educational Planning (RCEP),
UNESCO, Dubai, UAE
Anthony N. Haddad Amazon, Dubai, UAE
Victoria Heath Montreal AI Ethics Institute, Toronto, ON, Canada
ix
x LIST OF CONTRIBUTORS
Toufic Mezher Khalifa University of Science and Technology, Abu

Dhabi, UAE
Robert Mogielnicki The Arab Gulf States Institute, Washington, DC,
USA
Mohammed Sharfi Independent Researcher and Senior Consultant,
Ministry of Foreign Affairs, Doha, Qatar
Julia Singer Ludwig-Maximilians-University Munich, Munich, Germany
List of Figures
Chapter 3
Fig. 1 Methodology development 23
Fig. 2 Proposed solution framework 31
Chapter 4
Fig. 1 A brief timeline of AI and its main subfields, techniques,
and applications 49
Fig. 2 Singapore’s AI ecosystem 51
Fig. 3 The IT start-ups ecosystem in Bahrain in 2019 59
Fig. 4 The Wind Tower: A proposed model for constructing an AI
strategy for the Kingdom of Bahrain 60
Chapter 9
Fig. 1 The ABCDE model of Happiness Needs, as used in Dubai’s
Happiness Agenda (Al-Azzawi 2019). Reproduced
with permission from Smart Dubai 165
Fig. 2 Perspectives on experience with technology: People’s Needs,
User Context, Happy City 173
Fig. 3 Happy City Agenda tool published by the Global Council
for Happiness and Well-being (2019). Reproduced
with permission 174
xi
xii LIST OF FIGURES
Fig. 4 Mr Usta heatmap overlay on the city of Dubai, showing

areas of high demand for specific services. Reproduced
with permission 184
Fig. 5 As the capabilities of technology rise, the risk to humans
from technology is at a lower level than usually discussed.
Adapted from humanetech.com/problem (CHT 2019b) 189
Chapter 10
Fig. 1 World Economic Forum (WEF) Global Gender Gap Scores
for GCC Member States (2006), sourced from the WEF’s
2006 Global Gender Gap report 209
Fig. 2 World Economic Forum (WEF) Global Gender Gap Scores
for GCC Member States (2020), sourced from the WEF’s
2020 Global Gender Gap report 210
Fig. 3 Number of national/regional AI strategies per specificity
value for “Inclusion” policy area, sourced from CIFAR’s
Building an AI World (Kung et al. 2020) 228
Chapter 12
Fig. 1 Thematic overview of the analysed fatwas 289
Chapter 13
Fig. 1 Book “Wordcloud” 309
List of Tables
Chapter 3
Table 1 Resources of best practices in OGD 20
Table 2 Reported barriers to PSI 27
Table 3 Detailed example KPIs for measuring authorities’ responses
to data requests 34
Table 4 Proposed action timeline for the first 36 months of AD-DSI
strategy implementation 37
Chapter 6
Table 1 Descriptive Statistics of the Respondents’ Professional
Background 105
Table 2 Descriptive statistics of respondents’ experience
and familiarity with AI 106
Table 3 Knowledge and familiarity about AI 106
Table 4 ANOVA analysis of financial sector experience level 107
Table 5 ANOVA analysis of knowledge of AI 108
Table 6 Regression analysis 109
Table 7 Dependent variables on financial sector experience level
results 109
Table 8 Dependent variables on knowledge of AI results 109
Table 9 Summary of constructed hypotheses 111
xiii
xiv LIST OF TABLES
Chapter 8
Table 1 Sector breakdown of UAE free zones; the list is
not comprehensive and may include free zones
under construction 154
Chapter 10
Table 1 Overview of national and/or regional AI strategies
or guiding documents 218
Chapter 11
Table 1 Benefits and costs for users and companies 251
Chapter 12
Table 1 Analysed fatwas 288
PART I
Introduction
CHAPTER 1
An Introduction to AI in the GCC
1 Introduction
1.1 Background
Although the term Artificial Intelligence (AI) has appeared in literature as
early as the mid-1950s (Grudin 2008), there still is some confusion with
regard to its exact definition and context. The very early developments of
AI were driven by the gaming industry, particularly to develop a computer
program that could play chess (Newborn 2003). In 1997, nearly a half
century later, an IBM supercomputer called DEEP BLUE won in a
6-game chess match against Russian Grandmaster Gary Kasparov, making
the world chess champion the first to lose a match to a computer under
standard time controls, and demonstrating the success of AI and its ability
to outperform the human brain at a given task (Campbell et al. 2002).
E. Azar (B)
Khalifa University of Science and Technology, Abu Dhabi, UAE
e-mail: [email protected]
A. N. Haddad
Amazon, Dubai, UAE
© The Author(s), under exclusive license to Springer Nature 3

Singapore Pte Ltd. 2021
E. Azar and A. N. Haddad (eds.), Artificial Intelligence in the Gulf,
https://doi.org/10.1007/978-981-16-0771-4_1
4 E. AZAR AND A. N. HADDAD
Fast forward two decades, AI has now evolved from being driven by
the development of computer games to complicated systems that are
required for the functioning of our daily lives. As defined in a report
by PricewaterhouseCoopers (PwC) (PwC 2018), “AI is a collective term
for computer systems that can sense their environment, think, learn,
and take action in response to what they are sensing and their objec-
tives.” AI systems can have various levels of autonomy and adaptiveness,
ranging from systems that assist humans to perform tasks more efficiently
to ones with automated decision-making processes without any human
intervention (PwC 2018).
AI applications have gained a significant momentum in recent years
across various sectors such as health care (e.g., medical data analysis
and telemedicine), retail (e.g., product recommendations), manufacturing
(e.g., assembly line automation), infrastructure (e.g., smart cities and
self-driving cars), finance and banking (e.g., fraud detection), and educa-
tion (e.g., personalized learning).
1.2 Prospects
AI is a disruptive technology in today’s global economy. Beyond its
implications on how people live and work, it has tremendous economic
potential that remains untapped. With an incremental contribution esti-
mated at US$15.7 trillion by 2030, this is more than the output of China
and India combined (PwC 2017).
Given the nascent state of AI across countries, sectors, and individual
businesses worldwide, emerging-market actors may have a key opportu-
nity to leapfrog their more developed counterparts. The countries of the
Gulf Cooperative Council (GCC), led by the Kingdom of Saudi Arabia
(KSA) and the United Arab Emirates (UAE), have taken note. Compelled
by the need to diversify their economies away from oil, the GCC states
are actively pursuing knowledge-based economies. Moreover, the oppor-
tunity is significant, where among the GCC states, KSA is expected to
benefit the most from the move toward AI with an expected contribu-
tion of US$135.2 billion toward its economy. The UAE, in turn, will
benefit from a contribution of US$96.0 billion, while the remaining four
countries are expected to share a total growth of US$45.9 billion. In rela-
tive terms to the GDP of each country, the AI contribution to the UAE
economy will be the highest with 14% of its GDP (PwC 2018).
1 AN INTRODUCTION TO AI IN THE GCC 5
As a consequence, these countries have taken a progressive stance

toward AI investment and adoption. In the backdrop of its Saudi
Vision 2030 national transformation strategy, which has digital trans-
formation identified as a key enabler (KSA Government 2016), the
Kingdom famously granted Saudi citizenship to a female AI humanoid
robot (“Sophia”) at its first “Davos in the Desert” investment confer-
ence (Bloomberg 2018). More recently, in August 2019, it issued a royal
decree to establish a National Authority for Data and AI, and hosted in
October 2020 a virtual Global AI Summit (originally scheduled for May
2020 in Riyadh but delayed due to the coronavirus pandemic), touted to
be the largest AI meetup of its kind, under the patronage of Crown Prince
Mohammad bin Salman. Examples from the UAE include launching a
national AI strategy (UAE Government 2017), launching an autonomous
transportation strategy (Dubai Future Foundation 2017), appointing the
world’s first minister for AI (UAE Cabinet 2017), in addition to other
initiatives such as establishing an AI university—with full-fledged Masters
and PhD programs—based out of the country’s capital, Abu Dhabi
(Emirates News Agency 2019).
In response to the spread of the coronavirus, GCC countries have
found important practical applications in AI technologies to help slow
its spread. Enacting some of the world’s strictest measures, including
suspending passenger flights and imposing lockdowns with curfews, these
countries have also turned toward speed cameras, drones, and robots to
enforce social distancing and limited movement of citizens. In the UAE,
Dubai Police uses a program called “Oyoon” which through a network of
cameras in the city uses facial, voice, and license plate recognition to deter-
mine if a resident is employed in a vital sector or in possession of a valid
permit to leave their home or business (CNBC 2020). Additionally, the
city’s Road and Transport Authority (RTA) implemented computer vision
and machine learning algorithms to scan hundreds of thousands of hours
of video footage to detect and report violators of preventative measures
like proper wearing of masks, and proper distance between passenger
and driver (Emirates News Agency 2020). Government authorities have
also developed and heavily enforced the use of mobile phone apps, like
Tawakkalna and Tabaud in Saudi Arabia, and BeAware in Bahrain, to facil-
itate location-based contact tracing that monitors those who have tested
positive for the virus and to try to limit their exposure to the population
(The National 2020).
While headline-grabbing in nature, these initiatives are not isolated,

one-off events: They are part of serious attempts by the GCC states—led
by KSA and the UAE—to take a progressive and deliberate step toward
AI investment and adoption as an engine for growth. In so doing, they
seek to chart a path toward transitioning to a knowledge-based economy,
with the intent to compete with the world (and each other).
Yet in spite of this heightened level of GCC interest and investment
in AI, there are surprisingly no serious efforts to date to pull together
scholarship on the adoption challenges that exist, the implications on
society that these technologies may have, and the regulatory requirements
that need to be in place. Related efforts have typically been segregated
and remain limited in comparison with other growing economies such
as China and India. Important barriers are expected to hinder the full
deployment of AI in the GCC. These include—but are not limited to—
locally-generated knowledge, economic barriers, social risks, and rigid
institutional and policy structures. As such, important questions arise and
need to be further studied:
1. Are the Gulf countries on track to catch-up with the global push
toward AI?
2. Are the current physical and institutional infrastructures ready for
such transformation?
3. Is local talent available and ready?
4. Is the GCC economy ready for such job market evolution?
5. Will GCC countries remain adopters of AI knowledge and tech-
nology or will they evolve to become effective developers and
contributors at the local and national stages?
1.3 Objectives, Scope, and Target Audience

This book fills an important gap in the literature, with the first broad
reflection on the challenges, opportunities, and implications of AI in
the GCC. Unique results and insights are derived through case studies
from diverse disciplines, including engineering, policymaking and gover-
nance, economics, social science, and data science. Particularly related
to the “soft” science disciplines, we make some unexplored yet topical
contributions to the literature, with a focus on the GCC (but by no
means limited to it), including: AI and implications for women, Islamic
1 AN INTRODUCTION TO AI IN THE GCC 7
schools of thought on AI, and the power of AI to help deliver well-being

and happiness in cities and urban spaces. Finally, the book also provides
readers with a synthesis of ideas, lessons learned, and a path forward based
on the diverse content of the chapters. As such, we seek to have this be a
foundational text in what promises to be a growing area of research not
only in AI in the GCC, but of the contributions to reducing the existing
digital “North–South divide.”
The book caters to the educated non-specialist with interest in AI,
targeting a wide audience including professionals, academics, government
officials, policymakers, entrepreneurs, and non-governmental organiza-
tions. Given the gap in the literature on AI in the Gulf regional context,
the book is deliberately broad and diverse with chapters encompassing
multiple disciplines. As such, the volume is not intended to serve as a text-
book to undergraduate or graduate courses, though we see chapters or
parts (e.g., Part 3—Society, Utopia and Dystopia) being used in courses.
Finally, all of the chapters present original methods, results, and anal-
yses, which we anticipate will generate significant interest and citations in
future scholarly work on the topic.
As detailed in the following chapter, the book comprises 13 chap-
ters organized along three main themes pertaining to AI in the GCC:
(1) Data, Governance and Regulations; (2) Existing Opportunities and
Sectoral Applications; and (3) Society, Utopia and Dystopia.
The core of our contributions comes from a workshop held during
the annual Gulf Research Meeting (GRM), which took place on July
15–18, 2019 at Cambridge University in the United Kingdom. The
workshop was entitled “Artificial Intelligence in the Gulf Cooperative
Council (GCC) countries: Opportunities and Challenges” and comprised
of academics, practitioners, and policymakers of AI-related disciplines
from the six GCC countries (Saudi Arabia, United Arab Emirates, Kuwait,
Qatar, Bahrain, and Oman).
References
Bloomberg. (2018). Saudi Arabia Gives Citizenship to a Robot. News article.
https://www.bloomberg.com/news/articles/2017-10-26/saudi-arabia-
gives-citizenship-to-a-robot-claims-global-first.
Campbell, M., Hoane Jr., A. J., & Hsu, F. H. (2002). Deep Blue. Artificial
Intelligence, 134 (1–2), 57–83.
CNBC. (2020). Some Countries in the Middle East are Using Artificial Intelli-
gence to Fight the Coronavirus Pandemic. News article. https://www.cnbc.
com/2020/04/16/countries-in-the-middle-east-are-using-ai-to-fight-corona
virus.html.
Dubai Future Foundation. (2017). Dubai Autonomous Transportation Strategy.
Dubai Future Foundation, Dubai, UAE.
Emirates News Agency. (2019). Abu Dhabi Announces Establishment of the
Mohamed bin Zayed University of Artificial Intelligence. News article.
Emirates News Agency, Abu Dhabi, UAE. https://wam.ae/en/details/139
5302795116.
Emirates News Agency. (2020). RTA Employs AI in Taxis to Curb Spread of
COVID-19. News article. Emirates News Agency, Dubai, UAE. https://wam.
ae/en/details/1395302848858.
Grudin, J. (2008). A Moving Target: The Evolution of HCI. In The Human-
Computer Interaction Handbook—Fundamentals, Evolving Technologies, and
Emerging Applications. Lawrence Erlbaum Associates, Taylor & Francis
Group, New York, NY.
KSA Government. (2016). KSA Vision 2030. https://vision2030.gov.sa/en.
KSA Government, Riyadh, KSA.
The National. (2020). AI Helped Limit Spread of Covid-19 in the Gulf, Experts
Hear. News article. https://www.thenational.ae/uae/health/ai-helped-limit-
spread-of-covid-19-in-the-gulf-experts-hear-1.1063052.
Newborn, M. (2003). Deep Blue: An Artificial Intelligence Milestone. Springer,
New York, NY.
PricewaterhouseCoopers (PwC). 2017. Sizing the Prize: What’s the Real Value of
AI for your Business and How Can You Capitalize? PwC, London, UK.
PricewaterhouseCoopers (PwC). (2018). The Potential Impact of AI in the
Middle East. PwC, London, UK.
UAE Cabinet. (2017). His Excellency Omar Bin Sultan Al Olama. UAE Cabinet,
Abu Dhabi, UAE. https://uaecabinet.ae/en/details/cabinet-members/his-
excellency-omar-bin-sultan-al-olama.
UAE Government. (2017). UAE Strategy for Artificial Intelligence. UAE
Government, Abu Dhabi, UAE.
CHAPTER 2
Framework of Study and Book Organization
1 Book organization
The book consists of five main parts. Part I: Introduction—which is
comprised of two chapters—introduces the readers to the concept of Arti-
ficial Intelligence with a historical review both from the international and
local GCC contexts, followed by a mapping of the current gaps in the
literature motivating the need for the current volume. Part II: Data,
Governance & Regulations —which includes three chapters—presents case
studies on the supporting milieu needed to leverage advancements in the
context of multiple GCC cities. Part III: Existing Opportunities & Sectoral
Applications —which is composed of three chapters—covers case studies
and applications of AI initiatives in multiple sectors, such as finance,
economics and healthcare. Part IV: Society, Utopia and Dystopia—which
includes four chapters—offers a unique social perspective of AI, with
insights from studies on gender, religion, psychology and happiness. Part
E. Azar (B)
A. N. Haddad
Amazon, Dubai, UAE

VI: Conclusion—which presents an overview of the topics covered in

the various chapters—highlights key findings and directions for future
research. The following paragraphs detail each of the chapters presented
in the book starting from the chapters of Part II, which follow the current
introductory chapters.
2 Summary of chapters
In Chapter 3, “Public Sector Data for Academic Research: the case of the
UAE”, the authors explore the current procedures, barriers and possible
solutions for disclosing Public Sector Information (PSI) for academic
research in the UAE. The authors then propose a solutions framework
validated with public officials to better facilitate data sharing in the
country, which is an important prerequisite for research on data-driven
fields, such as AI.
In Chapter 4, “Strategy for Artificial Intelligence in Bahrain: Chal-
lenges and Opportunities ”, an AI strategy is presented for the Kingdom
of Bahrain following a multi-stakeholder analysis. The strategy is discussed
from different perspectives, including the availability of human capital and
the support from economic and financial institutions, while also aligning
it with the national strategy plans and priorities.
In Chapter 5, “Thoughts and Reflections on the Case of Qatar: Should
Artificial Intelligence be Regulated?”, the author tackles a fundamental
question about the role of government in regulating AI in the light of
the growing ethical, legal and security concerns of this technology. Based
on insights from the AI policy community in Qatar, the author calls for
a regulatory intervention from governments in order to strike a balance
between potential benefits and the expected threats from AI systems and
applications.
In Chapter 6, “Knowledge, Attitude, and Perceptions of Financial
Industry Employees toward AI in the GCC Region”, the authors inves-
tigate the knowledge, attitude and perceptions of professionals working
in the financial services industry in all six GCC countries. A statistical
analysis of data collected from 157 professionals shows an overwhelming
familiarity of the respondents with AI in the business and financial sector
context. The findings also highlight important concerns regarding ethical,
security and data privacy issues.
In Chapter 7, “The GCC and Global Health Diplomacy: The New Drive
towards Artificial Intelligence”, AI is studied in the context of its impact
2 FRAMEWORK OF STUDY AND BOOK ORGANIZATION 11
on the healthcare sector of GCC states and the emerging concept of

Global Health Diplomacy (GHD). The premise of this chapter is that the
financial resources provided by the GCC for AI applications in health-
care will positively contribute to GHD. In parallel, the need to engage in
global partnership and collaboration is highlighted to effectively leverage
the power of AI in the GCC health sectors.
In Chapter 8, “Free Zones in Dubai: Accelerators for Artificial Intel-
ligence in the Gulf ”, a unique analysis is presented on how economic
free zones can serve as useful vehicles for the effective implementation of
AI technologies in Gulf Arab economies. Taking Dubai as a case study,
the author argues that an alignment between AI innovation and ongoing
free zone developmental processes provides an amendable and sustainable
environment that maximizes the impact of AI services and applications.
In Chapter 9, “AI & Well-Being: Can AI make you happy in the city”,
the manifestation of AI technologies and applications is mapped to the
way that happiness and well-being are understood in the various contexts
of people’s lives. Drawing on examples from Dubai, and other global
initiatives, the author explores the utility of AI towards happier lives from
various perspectives, offering directions for further work to ensure a wider
coverage for all aspects of well-being.
In Chapter 10, “Women and the Fourth Industrial Revolution: An
Examination of the UAE’s National AI Strategy”, the participation of
women in the development, deployment and governance of AI is studied.
Using an extensive review of academic literature, news articles and data
from government and intergovernmental organizations, the author argues
that women’s inclusion in AI is a significant factor in ensuring the
successful development, deployment and governance of AI. A potential
strategy to include women in national AI strategies is also presented.
In Chapter 11, “The Art and Science of User Exploitation: AI in the
UAE and Beyond”, the authors explore the misuse of AI by Internet
companies to exploit vulnerabilities in human psychology and influ-
ence users’ views on matters ranging from political stands to promoting
product sales. The chapter provides unique insights on the neuroscience
behind the persuasion tactics used by these businesses, the strides and
challenges of AI use globally and in the Gulf region, as well as the
implications of AI across different disciplines.
In Chapter 12, “Fatwas from Islamweb.net on Robotics and Artificial
Intelligence”, the author evaluates the perceptions of robotics and AI
from an Islamic perspective. For that purpose, 14 Arabic and English
“Fatwas” are studied, which are legal opinions expressed by a Muslim

scholar or people with expertise in Islamic Law. The findings show that
scholars (1) have a fairly clear stance on the treatment of robotics but not
on AI; (2) they also do not show concerns with these technologies poten-
tially harming humans; and (3) avoid difficult issues such as the impacts
of developing strong AI.
Finally, in Chapter 13, “Outlook for the future of AI in the Gulf ”, the
book is concluded with an overview of the topics and case studies that
were covered in the various chapters. The chapter includes a discussion of
the key findings, insights learned, recommendations on how to address
the challenges towards a more effective implementation of and advance-
ment of AI technologies in the GCC, with directions for future research
on the topic.
PART II
Data, Governance and Regulations

CHAPTER 3
Public Sector Data for Academic Research:

The Case of the UAE
Aleksandar Abu Samra, Toufic Mezher, and Elie Azar
1 Introduction
Access to government data is one of the most important compo-
nents of economic and social development in society (Heeks, 2002). It
provides value to citizens and organizations, promotes public participa-
tion, improves the decision-making process, and fosters creative and inno-
vative solutions to contemporary problems (Janssen, 2011). More specif-
ically, in academic circles, acquiring information is critical to conducting
research and expanding scientific solutions (Mopas & Turnbull, 2011).
However, this process oftentimes faces many challenges, especially when
information lies behind the gates of government authorities. Academic
researchers often rely on their personal negotiation skills to get the infor-
mation needed because going through formal channels proves to be
hectic and time-consuming. Resistance to data sharing persists worldwide
because of institutional challenges rather than technical ones, “people”
A. Abu Samra · T. Mezher · E. Azar (B)


16 A. ABU SAMRA ET AL.
challenges (Harvey & Tulloch, 2006). The problems of hardly approach-

able and unresponsive authorities are widely recognized and actively
addressed by open government and transparency initiatives, but the
situation varies significantly between different countries.
Thanks to rapid technological improvements, data on many different
types of information has become abundant, easy to collect, and inexpen-
sive to store (Lane et al., 2014). As key players in the infastructure sectors
are typically government entities and public-private partnerships, relevant
data naturally falls under the public domain. Acknowledging the value of
collected information, but also limitations to coping with its vastness, it
is not surprising that governments around the world are introducing new
data-sharing policies to engage other stakeholders toward understanding
and getting insights out of it (Einav & Levin, 2013). Freedom of Infor-
mation Acts and Open Data movements are on the rise, and now more
than ever, datasets are possible to combine across borders and disciplines,
further accelerating the societal progress.
In the United Arab Emirates (UAE), recent investments into higher
education hubs and a rapidly evolving research ecosystem aim to place
the country as a global leader in research and innovation (Byat & Sultan,
2014; Al-Hammadi et al., 2010), particularly in the field of Artificial
Intelligence (AI). This is confirmed by the number and scale of AI initia-
tives that are being undertaken, such as the launching of the national
AI strategy or the new Mohamed bin Zayed University of Artificial
Intelligence (UAE Government, 2017; Emirates News Agency, 2019).
However, despite these advancements, data sharing policies remain
limited both in terms of Open Data (World Wide Web Foundation,
2015), and more rudimental Right to Information laws (World Justice
Project, 2015). For this reason, access to Public Sector Information
(PSI) is burdensome to scientists, a condition that was shown to signifi-
cantly slow down innovation and the research potential in many countries
(Alani et al., 2007; Arzberger et al., 2004; Commission of the European
Communities, 1999).
There is a great need to promote data sharing and collaboration
between government and academia in the UAE, especially for data-
intensive fields, such as AI. Although there are many studies, globally and
from the region, that assess the state of openness through international
rankings, there is a lack of studies exploring this problem in more depth
and focusing on specific challenges and potential solutions.
3 PUBLIC SECTOR DATA FOR ACADEMIC … 17
The main goal of this chapter is to comprehensively evaluate and

understand the current state of sharing PSI in the UAE, as well as propose
feasible solutions for improving it. For achieving that goal, objectives were
split into the following segments:
– Understand what logistical, communicational, legal, and personal

challenges researchers encounter in the process of obtaining PSI
necessary for their studies.
– Understand what protocols are followed by the UAE authorities
when approached with data requests and detect the key barriers to
sharing government data with third-parties, particularly academia.
– Understand the latest international practices used to promote the
sharing of PSI and possible implementation challenges in the UAE.
– Propose recommendations to modify international practices to
address locally discovered challenges and broader understanding of
the local context.
– Combine the developed recommendations to propose a solution
framework consisting of policies, strategy, and action timeline to
introduce improved data sharing in the UAE.
– Validate proposed solutions with local authorities and experts in the
field.
2 Literature Review
2.1 Challenges to Data Sharing
Challenges to freedom of information are certainly very common in
both developed and developing parts of the world (Banisar, 2006), with
notable debates happening in the fields of access to citizen data, ethics of
data sharing, and in the regulatory fields (Cohen et al., 2014; LAPSI 2.0,
2014).
From a user’s perspective, challenges to government information access
vary by demographics and were shown to be experienced in Research
and Development (R&D) (Beniston et al., 2012; Mopas & Turnbull,
2011), by journalists (Garrison, 2000), in inter-organizational sharing
(Fedorowicz et al., 2010), and at the international scale (Harris &
Browning, 2013).
Sharing knowledge between individuals and organizations is a complex

issue, influenced by anything from personal and psychological incen-
tives to institutional procedures and laws (Yang & Maxwell, 2011). At
individual level, information can be perceived as property and therefore
becomes an asset worth protecting for increasing one’s status and influ-
ence, or used as a tool for securing personal power inside organizations
(Ardichvili et al., 2003; Constant et al., 1994; Marks et al., 2008; Willem
& Buelens, 2007). Across organizations, barriers to sharing information
can be the time and effort needed to provide it, as well as exposure to crit-
icism, especially if done without receiving proper recognition (Ardichvili
et al., 2003; Cress et al., 2006). Furthermore, due to limited resources,
organizations naturally prioritize immediate issues over long-term bene-
fits to information exchange (Landsbergen & Wolken, 2001; Zhang
& Dawes, 2006). Inter-organizational sharing of knowledge encoun-
ters further problems due to different origins, values, and cultures of
participating organizations, often invoking complexity, misunderstanding,
distrust, or even competition (Drake et al., 2004).
2.2 Open Data and Open Government Data (OGD)

Open Data, defined as data that is “free to use, modify, and share by
anyone for any purpose” (Open Knowledge Foundation, 2015), enjoys
increased attention going hand in hand with the rapid development
of technology and information distribution. Government authorities,
although in possession of much-valued data on society, do not have
resources to process it alone, and need to outsource the initiation
of societal challenges to their citizens (Eckartz et al., 2014). When
addressing strictly PSI, the term Open Government Data (OGD) is
commonly used, annotated by eight original principles stating the data
should be complete, granular, timely, accessible, machine-processable,
non-discriminatory, non-proprietary, and license-free (Tauberer, 2007).
Still, Open Data has many challenges on its own. Publishing structured
datasets is new for local governments and, therefore, causes confusion
on how data should be released (Conradie & Choenni, 2014). Some of
the dominant barriers include data quality, ownership and trust issues,
privacy, and economic impediments. To unlock its potential, data needs
to be governed (Cheong & Chang, 2007; Weill & Ross, 2004), but there
is a lack of tools for sharing, data definitions are conflicting, and privacy
concerns are many (Zhang et al., 2005). Practical frameworks have been
developed to guide through these barriers, but on the local levels, there is
oftentimes an absence of understanding of the added value of such change
(Kassen, 2013). Table 1 presents an extended list of the most relevant
documents used for understanding international practices in this study.
2.3 Information Sharing Trends in UAE

The UAE national goals are clearly set on the direction toward a
knowledge-based economy and society. The UAE Vision 2021 (Prime
Minister’s Office, 2010) announced a national era of competitive knowl-
edge economy based on entrepreneurship and innovation. Furthermore,
the Abu Dhabi Department of Economic Development (2009) has
acknowledged that the knowledge-based economy is not merely a new
branch of economy, but rather an entire transition to an economy based
on information. However, as data is the key driver of such change, it
is important to address the current gaps to harnessing its full potential.
In the latest Open Data Barometer report (World Wide Web Foun-
dation, 2015), the country scored low compared to its broader state
of prosperity. An equivalent to a “Freedom of Information” act is still
not available, and the UAE generally ranks low in Open Government
Index by the World Justice Project (2015). Finally, a yearly review of
datasets published by government agencies in the UAE carried out by
the Global Open Data Index (2015) ranked the UAE at the bottom 20%
of all reviewed countries concerning various information availability and
structure criteria.
In recent years, Dubai promised to bring positive progress to this area
(Government of Dubai Media Office, 2015). However, few studies have
already tackled the current state of data “openness” in the UAE, looking
at information present on the official government Web sites (AlAnazi
& Chatfield, 2012) or directly interviewing the authorities (Elbadawi,
2012). Still, little research efforts have been conducted on problems that
emerge when governmental data is directly requested for academic use,
exploring issues of the current government–academia relationship over
data in the UAE, or on how international trends can be applied to the
local context.
This chapter aims at closing this gap by interviewing both academicians
and public officials in the UAE in order to investigate how govern-
ment authorities approach data requests by academia, and what are their
perceptions of the trending shifts toward Open Data. A proposed solution
20
Table 1 Resources of best practices in OGD
Source Type Reference
International Organizations Open Government Data Toolkit (World Bank, 2016)

Open Data Handbook (Open Knowledge Foundation, 2016)
Open Data: Measuring What Matters (Hadjigeorge, 2016)
Open Data Barometer Global Report (World Wide Web Foundation, 2015)
International Open Data Charter (Group of Eight, 2015)
Guidelines for Open Data Policies (Sunlight Foundation, 2014)
A. ABU SAMRA ET AL.
Federal Governments Open Data Policy—Managing Information as an Asset (Project Open Data, 2016)
Open Data in the G8: A Review of Progress on the Open Data Charter (Castro & Korte,
2015)
Supporting the National Information Infrastructure (Public Data Group, 2014)
Open Data Strategy 2014–2016 (Department for Business Innovation & Skills, 2014)
Canada’s Action Plan on Open Government 2014–2016 (Treasury Board of Canada
Secretariat, 2014)
Open Government Plan (General Services Administration, 2012)
City and County Governments LA Open Data Policy and Playbook (Los Angeles Data Team, 2015)
Open Data Guide by the City of Philadelphia (Headd, 2014)
Open Data in San Francisco: Institutionalizing an Initiative (Bonaguro, 2014)
Open Data Implementation Plan (Montgomery County Government, 2014)
Open Data Policy—Implementation Plan (City of San Diego, 2014)
Open Data Policy and Technical Standards Manual (Bloomberg & Merchant, 2012)
NYC OpenData Technical Standards Manual (NYC OpenData, 2012)
Source Type Reference
Research Papers and Case Studies Planning and designing open government data programs: An ecosystem approach (Dawes
et al., 2016)
A systematic review of open government data initiatives (Attard et al., 2015)
Open Data in the Legislature: The Case of São Paulo City Council (Matheus & Ribeiro,
2014)
Open Data Ireland: Best Practice Handbook (Lee et al., 2014)
Open Government Data: Towards Empirical Analysis of Open Government Data Initiatives
(Ubaldi, 2013)
Open data: Emerging trends, issues and best practices-a research project about openness of
public data in EU local administration (Fioretti, 2012)
Open data: an international comparison of strategies (Huijboom et al., 2011)
3
PUBLIC SECTOR DATA FOR ACADEMIC …
21
framework is then presented to guide efforts toward implementing such

trends while staying consistent with the current regulatory limitations and
practices in place.
3 Methodology
The methodology used for conducting the interviews in this study was
split into five phases: (1) interview design (2) data collection (3) data
analysis and identification of barriers (4) understanding international prac-
tices and their challenges, and (5) proposing a comprehensive solution
framework. Figure 1 visualizes details of this methodology:
3.1 Interview Design

The aim of the interview design phase was to provide a tool to collect
information about the process of obtaining datasets from the public sector
in the UAE, from both the perspective of academicians (who request the
datasets) and public officials (responsible for processing those requests).
To capture the full spectrum of subjects’ experiences, a face-to-face qual-
itative interview approach was used following best practices in interview
protocols (Seidman, 2013; Turner III, 2010; Weiss, 1994).
The interview with academicians had the objective of identifying
logistical, communicational, legal, personal, and any other challenges
encountered in obtaining PSI needed in their studies. The interview also
explored the nature of the PSI requested, as well as personal opinions
on how the researchers thought the system might be improved. The
interview was split into four sections:
1. Getting to Know the Interviewee—To briefly give a context of the

subject’s background and ongoing work.
2. Project-Related Details —Aimed at understanding why subjects need
data from public authorities, the complete process of requesting it,
and the related barriers in obtaining it.
3. Thoughts on Possible Improvements —Subjects are asked to give their
opinions on the current state of data availability in the UAE, and
steps toward improving it.
4. Closing the Interview—Lastly, the subjects are asked to refer to other
people that experienced similar problems and give general feedback
on the interview process.
Fig. 1 Methodology development
The interview with public officials had the objectives of understanding

protocols followed by the UAE authorities when approached with data
requests from academia, the concerns these requests might cause, and
personal opinions on how these concerns could be mitigated. The
interview was split into five sections:
1. Getting to Know the Interviewee—A brief introduction to the

subject’s position in his or her organization.
2. Data Collection Processes —Understanding what type of data the

authority collects or possesses, the formats used, and their access
hierarchy.
3. Infrastructure for Open Data—Understanding familiarity with Open
Data, current data sharing protocols in place, and barriers to a more
open approach.
4. Thoughts on Openness —Collecting opinions on possible improve-
ments in data sharing practices from both government and from
academia sides.
5. Closing the Interview—Final thoughts, feedback from the inter-
viewee, and references for further interviews.
After developing the first versions of both interviews, test-runs were

performed with subjects that agreed to give in-depth feedback. As a
result, some questions were shortened and made clearer, while some were
identified as redundant and therefore removed.
3.2 Data Collection and Scope of Work

Twenty-two researchers in academia were interviewed across ten research
institutions across the Emirates of Abu Dhabi, Dubai, Sharjah, and
Ajman. This included 17 faculty members, four graduate students (three
Doctoral and one Master’s student), and one research engineer. Three
participants were three females and 19 males. Initial subjects were iden-
tified by word of mouth recommendation from experts in the field,
followed by referrals acquired by explicitly asking subjects for further
contact suggestions. Interviews were scheduled by sending a short intro-
ductory email directly, asking for a 20-minute interview benefiting the
study. If the respondents seemed to be busy, they kept things to the point
and within the announced interview time, but if the respondents had extra
time and wanted to contribute more, interviews continued for up to an
hour.
Twelve people were interviewed across nine public authorities in the
UAE and one consultancy company involved in local e-government
efforts. Contact details of potential subjects were initially acquired by
reaching their respective departments through information found online,
after which a call, email, or snail-mail letter to the department would
follow. Later on, the “snowball sampling” approach was used, where the
investigator asked already interviewed subjects to refer him to relevant
people in other public authorities. The investigator would then email or

call the new contacts, briefly introduce the study, and ask to schedule an
interview.
3.3 Data Analysis and Identification of Barriers

In analyzing notes collected during the interviews, a general inductive
approach was used for analyzing qualitative data (Thomas, 2006). In the
results section, findings are presented as objectively as possible, basing
them solely on interviewees’ responses. Key barriers to data sharing were
then identified from both the perspectives of academic researchers and
government officials.
3.4 Understand Challenges to Implementing Best Practices

This phase of the methodology first consists of thoroughly reviewing
best practices in Open Government Data (OGD) from a multitude of
sources including international Open Data organizations, policy docu-
ments and guidebooks from federal and city governments, as well as
research papers and case studies reviewing and analyzing different OGD
strategies. After the practices had been properly understood, recognized
patterns were split into separate Best Practices for easier further analysis.
Each major recognized Best Practice was related to the context of UAE
culture and procedures in place, as well as the previous findings on the
existing key barriers to data sharing. Arguments were made on why, if the
UAE was to follow such international practices directly, challenges could
be encountered and the progress potentially stalled.
3.5 Propose Solution Framework

Based on each recognized Best Practice of Open Data, and for every
recognized challenge to its implementation in the UAE, a specific recom-
mendation was proposed to address it. The adapted Best Practices were
finally assembled in a solution framework, proposing how such a modified
approach to Open Data could be successfully implemented in the Emirate
of Abu Dhabi. The developed solution framework, which is detailed in the
upcoming sections, consists of four layers; each layer incorporated rele-
vant Best Practices and their respective developed recommendations, while
again referring to barriers identified from the previously held interviews:
1. Strategy—Describing the proposed policy and actions performed by

the entities responsible for its development;
2. Execution—Describing actions of the government entity proposed
to put the developed strategy into action;
3. Operation—Describing responsibilities of government authorities
that would participate in this solution by providing datasets to the
system;
4. Process —Describing how the solution works from a user perspective.
To validate such a solution framework, the author held meetings with

policy experts and people in the UAE government authority, some of
whom also participated in the previous interviews. Based on the given
feedback, and for each suggestion relevant to the scope of the study,
recommendations were updated, and the framework reassembled. After
the saturation point was reached, the proposed solution framework was
considered to be in its final version.
4 Results and Analysis

4.1 Barriers to Accessing Public Sector Information
From the perspective of interviewed researchers, feelings about the
current government–academia relationship in the UAE were mixed. In
many cases, the researchers reported the willingness of public officials to
help, but the attitude was never enough to lead to successful outcomes.
Obstacles commonly faced by researchers include identifying who has the
data and obtaining the requested information in a timely manner and
in the right format. Lack of relevant laws and regulations caused data
requests to appear unusual, requiring multiple in-person meetings with
government officials to justify such requests. Furthermore, the researchers
had to be affiliated with institutions that had pre-signed Memorandums
of Understanding (MoUs) with respective government bodies, and finally,
the researchers would have to sign Non-Disclosure Agreements (NDAs),
preventing them from sharing data even with their colleagues.
The interviews within public authorities in the UAE revealed the
complementing side of the narrative: The UAE authorities indeed have
good intentions in supporting local academia and research, but the
processes to attain such intentions with successful data exchange are
unclear, and many obstacles exist toward a faster and more open approach.
Table 2 Reported barriers to PSI
Academia’s perspective Public authorities’ perspective
Identifying Authorities with Needed Data Absence of Data Sharing Laws and
Regulations
Finding Contacts and Initiating Concerns about Quality, Completeness,
Communication and Availability of Data
Lack of Procedures Security and Privacy Concerns
Paperwork Misuse and Misinterpretation Concerns
Slow Response by Authorities Lack of Incentives to Share Data
Unstructured and Undocumented Data
Unfamiliarity with MoUs and NDAs
The perception of interviewed respondents was that the current systems

in place performed well. Obstacles to a more open approach to data
sharing are many and include lack of laws and legislations, concerns about
the quality of shared data, but also concerns on how the shared data
would be used and represented. Employees within public authorities see
no immediate benefits to sharing data and therefore lack direct incen-
tives to optimize their processes. In summary, Table 2 presents the most
commonly reported barriers by the interviewees.
4.2 Adapting International Best Practices

The following are summaries of main Best Practices of implementing
Open Data on a government level, selected from subjectively recognized
patterns in the international practices referenced in Table 1. In order to
highlight the differences between the UAE and the countries who devel-
oped these best practices, each of the Best Practices is carefully studied,
identifying its possible challenges to implementation in the local context.
Finally, specific recommendations were developed to modify each Best
Practice for successfully solving such challenges.
4.2.1 Best Practice #1: Data Sharing Principles

In short, Open Data is defined as: “Data and content that can be freely
used, modified, and shared by anyone for any purpose” (Open Knowl-
edge Foundation, 2015). Together with the eight underlying principles
of OGD (Tauberer, 2007), Open Data is not fully compatible with the
current practices of government authorities in the UAE. In the absence
of laws on rights to information access, the UAE public authorities feel a

significantly larger degree of direct “ownership” for such collected infor-
mation. For a successful solution that would facilitate improved access
to PSI, the political and cultural realities must first be fully understood.
As the main issues of accessing PSI by researchers in the UAE are of
communicational and procedural nature, local authorities should focus
on improving access to data that exists already, in the form that is already
kept, with permissions and usage rights that are already in place.
4.2.2 Best Practice #2: Government Structure Implementation

International practices to implementing OGD typically rely on appointing
new, dedicated government bodies for its plan and execution. To avoid
the logistical complexity and unnecessary risk such approach might cause,
Abu Dhabi can decide for a more integrated approach: Abu Dhabi
Systems & Information Centre (ADSIC) is leading the current infor-
mation sharing in the capital Emirate, with a mission to “[…] Deliver
innovative digital services, standards and policies, while building the
foundation for an ICT -Mature society” (ADSIC, 2016). ADSIC has
already implemented the Abu Dhabi Spatial Data Infrastructure (AD-
SDI) program that focuses on sharing geospatial data among numerous
public stakeholders in the Emirate. The government-to-government data
communication they provide uses modern, sophisticated systems, with
MoUs already in place. More than any currently existing public authority
in Abu Dhabi, ADSIC has the know-how, the experience, and the govern-
ment support to facilitate data exchange between different stakeholders
of PSI. For those reasons, ADSIC could potentially form another internal
body for executing the newly introduced data sharing practices.
4.2.3 Best Practice #3: Identifying Publishable Datasets

Various international practices for identifying publishable datasets focus
on transparency of internal government processes, especially in coun-
tries funded by taxpayer’s money; the emphasis on such information in
the UAE is expected to cause unnecessary implementation difficulties.
Therefore, the guidelines for deciding on datasets valuable for publishing
should first exclude any data related to government expenses, internal
processes, people accountable, or other similar information. Efforts saved
by not publishing such data can largely be used for focusing on datasets
of direct value to the local stakeholders. More specifically, combining
approaches mentioned in the international practices, authorities should
focus on publishing data directly benefiting research, innovation, and

economy in the country.
4.2.4 Best Practice #4: Prioritization Criteria

Similarly, to Best Practice #3, international OGD practices often advise
authorities to prioritize datasets that would increase transparency and
accountability of their internal decisions and processes. Given the barriers
listed in Table 1, priority cannot be fully put on the transparency issues,
risking rejection by both policymakers and authorities themselves. As a
compromise solution, this study recommends prioritizing the publication
of data according to the UAE national goals and future visions.
4.2.5 Best Practice #5: Data Portal and Access Rights

A common practice included in OGD solutions is having a unifying web
portal that serves as a structured, categorized, and searchable database of
all “public” information published by relevant government authorities to
date. Taking into account the current strict criteria for perceiving data as
“public,” the majority of the datasets would still fall under the “restricted
public,” or “non-public” category, therefore unavailable through the web
platform provided. Consequently, a newly created portal must not focus
solely on the collection and categorization of datasets (as typical Open
Data portals do) but instead serve as a platform to facilitate communica-
tion between people and organizations in need of data on one side, and
public authorities on the other. Because many of the datasets owned by
the UAE authorities are considered “restricted,” the portal should also
route requests for such data to relevant authorities to resolve.
4.2.6 Best Practice #6: Phasing Out Implementation

A typical timeline for international OGD implementations often includes
direct steps toward satisfying the end goal of any such implementation;
such steps can be announcing an Open Data Policy, appointing an Open
Data Board and a Chief Data Officer, or developing a unifying Open
Data Portal. These implementation practices, although reasonable and
not overly ambitious, would not be suitable for the context of UAE
for one main reason: Such order of developments does not immedi-
ately address the most critical challenges of data access in the country,
which is facilitation communication between data users and public author-
ities. To immediately address the critical challenges of PSI access in Abu
Dhabi, a three-phase approach is therefore recommended, as detailed in

Sect. 4.3.3.
4.2.7 Best Practice #7: Engaging Stakeholders

To catalyze the use of published data, international OGD practices focus
on stakeholder engagement with events targeting the general public. A
major challenge to following such practice in the UAE is the assumption
that the majority of datasets will still be restricted to only academia and
other important stakeholders with special permission to use it. Conse-
quentially, a more optimal approach would be to focus engagement
efforts on important stakeholders, as currently done by ADSIC, proposed
in Best Practice #2.
4.2.8 Best Practice #8: KPIs and Accountability Practices

KPIs measuring the performance of international OGD efforts mainly
fall into three categories: (1) activity metrics (2) quality metrics, and
(3) impact metrics. Such developed metrics do not measure the current
most important challenges in the UAE—the general responsiveness of
government authorities to data requests. As the major complaints in
the current process of obtaining datasets from public authorities are of
communicational and procedural nature, in addition, the metrics that
target Quantity, Quality, and Impact of data, Timeliness of responses to
data requests should be not only introduced but prioritized. Taking the
time needed for successfully processing each request is critical for incen-
tivizing authorities toward a more dedicated performance. Moreover, the
metrics on Quantity and Quality should also include metrics specialized
in tracking the performance of authorities’ responses to data requests.
4.3 Proposed Comprehensive Solution Framework

A macro-sketch of the proposed comprehensive framework is provided in
Fig. 2 for easier visualization and communication of proposed ideas to
the reader. The framework combines developed Best Practices of relevant
international practices, respective recommendations on how they can be
better implemented in the UAE, and currently existing barriers identified
after analyzing the performed interviews. On the left side of Fig. 2 are
listed brief summaries of the investigated challenges to data sharing. Each
is marked with a flag number to indicate what elements of the proposed
framework play a role in addressing the existing problems. Within the
Fig. 2 Proposed solution framework

diagram, four layers of the implementation are marked with dashed lines
to distinguish Strategy, Execution, Operation, and Process layers. White
rectangles represent existing government entities or bodies. Outlined in
red are newly proposed structures or protocol elements. Colored rect-
angles contain numbers to indicate how do the developed Best Practices
and their respective adjusting recommendations fit in the big picture of
the proposed framework (e.g., Best Practice #1: Data Sharing Principles ).
4.3.1 Strategy
First and foremost, a policy directive at the level of the Emirate of Abu
Dhabi needs to be initiated by the Abu Dhabi Executive Council. The
Executive Council would, therefore, announce a Data Sharing Policy &
Strategy to state the willingness of Abu Dhabi government to implement
the new data-sharing infrastructure, as well as provide general guidelines
for its execution. Following the arguments from Best Practice #1, such
a policy should take a relaxed approach to introduce data sharing prin-
ciples, focusing on improving access to already existing datasets, in their
currently present forms, with permissions and usage rights that are already
in place. A detailed strategy to fulfill those principles should follow,
providing know-how for Authorities on how to identify which datasets
are suitable for publication, as well as how to prioritize their release, as
explained in Best Practices #3 and #4, respectively.
In the future, the policies and strategies could also be responsibilities
of a UAE Federal Level Data Sharing Infrastructure, if such happens to
be developed. By introducing the Data Sharing Policy & Strategy, the
following currently existing challenges are expected to be resolved or
mitigated:
– Absence of Laws and Regulations
4.3.2 Execution
For the reasons explained in Best Practice #2, Abu Dhabi Executive
Council shall appoint ADSIC to execute the implementation of the new
data-sharing infrastructure. Using the know-how, signed agreements, and
processes in place for their existing AD-SDI program, ADSIC shall
create a new internal body—Abu Dhabi Data Sharing Infrastructure
(AD-DSI)—to facilitate data sharing inside the Emirate. AD-DSI would
be focused on fulfilling the principles described in Best Practice #1 by
further developing Data Sharing Policy & Strategy and coordinating its
execution with public Authorities. Following the set Strategy, AD-DSI
would standardize the common structuring and documenting practices,
educating Authorities on how to prepare their datasets for faster and easier
use by Data Consumers. In addition, AD-DSI would be responsible for
creating and managing the Abu Dhabi Data Sharing Portal (AD-DSP),
also referred to as the Portal.
Once the system is operational, AD-DSI would start regularly
receiving data and data dictionaries from public Authorities, managing
them, and sorting for display on the Portal. At the same time, through
the Portal, AD-DSI would receive data sharing requests from Data
Consumers. Such requests would either be handled directly by AD-DSI
if the data is available and all permissions acquired, or if not—routed
to relevant Authorities for further processing. It is also recommended
that ADSIC expands its Stakeholder engagement to people and entities
involved in contribution to or usage of the Portal. This can provide
ADSIC with insights and directions on how to best cater to its Stake-
holders’ needs and concerns, as further explained in Best Practice #7 .
By placing a central body to coordinate Authorities in their data
sharing and structuring practices, the following currently existing chal-
lenges are expected to be resolved or mitigated:
– Unstructured and Undocumented Data

– Concerns About Quality, Completeness, and Availability of Data
4.3.3 Operation
Local Authorities would, under the supervision of AD-DSI , execute the
following main responsibilities: Produce and maintain data dictionaries —
digital inventories that account for all data assets created or collected by
the Authority; classify their datasets according to different access rights;
respond to data sharing requests routed through AD-DSP, as well as
proactively provide datasets to AD-DSI ; and measure and report their
progress based on introduced KPIs.
Producing data dictionaries would be the first step in the process
towards a more structured and efficient approach to data sharing by each
Authority. By having such inventories of all data assets already in posses-
sion, Authorities will have an overview of the amount, complexity, state,
and form of available data to help them make a strategy for its further
classification and publication.
A four-layer data classification approach is proposed based on Best Prac-
tice #5. Public data would be the data freely available to access and use
by anyone for any purpose. It would be hosted on the Portal or listed
for a release by request. Restricted data would be accessible only by Data
Consumers whose host institutions have signed appropriate data sharing
MoU with ADSIC. An individual might or might not need to sign an
already prepared NDA before such data is used or processed as a publi-
cation. Sensitive data is data with very strict usage limitations that, on
top of signing a specifically assembled NDA, require in-person meetings
with the authority that owns it before a decision to disclose the requested
information is made. Confidential data is not meant to be released for any
reason and for any entity.
Finally, proposed in Table 3 are detailed examples of metrics suit-
able for tracking the performance of each Authority, assembled by
using relevant metrics suggested by Hadjigeorge (2016) and Bonaguro
Table 3 Detailed example KPIs for measuring authorities’ responses to data

requests
Category Example KPI
Quantity Percent of inventoried datasets that are published

Percent of inventoried datasets published by data classification type
Percent of inventoried datasets published by priority level
Percent of requested datasets publisheda
Quality Percent of datasets updated with required metadata
Percent of repeated requests due to the low quality of initially released
dataa
Percent of repeated requests due to the misunderstanding of the initial
request made*
Impact Number of respondents indicating that open data has improved their
analytical work
Number of products or publications made with shared data
Number of data requests by data classification typea
Timeliness Response time satisfaction ratea
Average time needed for initial response to each requesta
Average time needed for closing successful requestsa
Average time needed for closing unsuccessful requestsa
Number of meetings held with people that requested dataa
a Newly proposed KPIs to track Timeliness and success rate of responses to data requests
Another random document with
no related content on Scribd:
Elizondo and cross the upper Bidassoa. It is scarcely credible that the
men who had fought for ten hours under such conditions on such
rough ground, retained strength to move another furlong—but the
order was obeyed, though many badly wounded men had to be
abandoned, and though the chaussée was strewn for miles by dead-
beat stragglers, who dropped out and slept till daylight. They were not
disturbed—for D’Erlon made no move till the sun was well up—he
had won the pass and was expecting to have to fight again at dawn,
for the right to emerge from it.
The losses in a fight so honourable to the British battalions, if so
discreditable to British generalship, had been immense in Cameron’s
brigade, heavy in Pringle’s, appreciable among Barnes’s men, who
only struck in at the eleventh hour. The first-named had lost 800 men
out of 1,900 present, of whom 343 belonged to the gallant and
unlucky 1/92nd. Pringle’s three battalions had 530 casualties out of
2,000 present, including 140 unwounded prisoners taken on the
Gorospil knoll from its light companies. Barnes and the 7th Division
troops had won a glorious success with a loss of only 140 men. The
total list gives just under 1,500 casualties out of 6,000 men engaged
—of whom 349 were prisoners (200 of them wounded). These are
very different figures from the 3,500 total at which D’Erlon stated
Stewart’s loss—but sufficiently distressing. The enemy had suffered
still more, but from infinitely greater numbers—their commander
reported 1,400 casualties in Darmagnac’s division out of 7,000
present, 600 in Maransin’s. Abbé was barely engaged at the eleventh
hour: one of his brigadiers (Rignoux) was hit and only four other
officers, with perhaps 100 men. The total reported is therefore about
2,100—no very formidable proportion out of 20,000 men present. But
some battalions had been badly cut up—the 103rd of Maransin’s
division, which bore the first fury of Barnes’s attack, had 15 officers
killed and wounded out of 20 present; and the 28th of Darmagnac’s
division lost a similar number in sustaining the attack of the right wing
of the 1/92nd and the British 28th. But this was a two-battalion
regiment with 40 officers present. Nevertheless, D’Erlon’s report to
Soult sings victory in very modest terms—he has captured the
enemy’s position and holds it at the end of the day—the affair had
been one of the most desperate that he has ever seen—the enemy’s
loss has been far greater than his own—but there is no blowing of
trumpets.
SECTION XXXVIII: CHAPTER IV
SORAUREN. JULY 26-28
The first day’s fighting in the Pyrenean passes could not be called
satisfactory either to Wellington or to Soult. The former had lost both
the defiles in which he had intended to make his first stand, and had
lost them in a very tiresome fashion—he thought that Maya might
have been held at least for twenty-four hours, if there had been a
divisional general on the spot to direct the defence: while
Roncesvalles had not been forced, but abandoned by Cole, who
could certainly have made a longer resistance, if only the orders sent
to him had been obeyed. It was, above all things, necessary to gain
time for the concentration of the army, and a precious day had been
lost—and need not have been lost.
But Soult can have been no better pleased: time, to him also,
meant everything; and the orders which he had issued to his
lieutenants had presupposed an easy triumph by surprise in the early
morning, with a forward march in the afternoon. Instead of this he had
won by nightfall a bare foothold on the summit of each pass, after
much fighting of an unsatisfactory sort. He, too, had lost a day; and it
was only on the morrow that he discovered that both at Maya and at
Roncesvalles the enemy had slipped away in the dark, leaving to him
the power to debouch from the defiles.
BATTLE OF SORAUREN
July 28th 1813 Showing the General Situation at 1.15 PM.
Nevertheless, the Marshal sent on the morning of the 26th a very
flamboyant message of victory to his master the Emperor, who then
lay at Mayence. Both Maya and Roncesvalles had been forced,
D’Erlon had captured five guns and many hundreds of prisoners at
the former pass: he himself hoped to be at Pampeluna, and to have
raised its siege, by the 27th. These news were sent on from Bayonne
by semaphore to Paris and the Rhine, and reached Napoleon on
August 1st. At the same time, and by the same rapid method of
transmission, arrived General Rey’s report of his successful repulse
of the assault of July 25th. It is worth while to turn away from solid
history for a moment, in order to see how the Imperial editor of the
Moniteur utilized this useful material for propaganda. He first wrote to
Clarke the Minister of War: ‘We can now give the public some
account of affairs in Spain. The Vittoria business and the King must
not be mentioned. The first note which you must put in the Moniteur
should run as follows—“His Majesty has named the Duke of Dalmatia
as his lieutenant-general commanding his armies in Spain. The
Marshal took up the command on July 12, and made immediate
dispositions for marching against the besiegers of Pampeluna and St.
Sebastian.” After that put in General Rey’s first letter about the events
of the 25th-27th. You had better make some small additions to the
number of prisoners and of guns captured, not for French
consumption but to influence European opinion. As I am printing
General Rey’s dispatch in the Frankfort Journal, and have made
some changes of this sort in it, I send you a corrected copy so that it
may appear in the Moniteur in identical terms.’
The Emperor’s second letter to his Foreign Minister, the Duke of
Bassano, sent from Dresden three days later, is even more amusing.
‘You had better circulate the news that in consequence of Marshal
Soult’s victory over the English on July 25, the siege of St. Sebastian
has been raised, and 30 siege-guns and 200 waggons taken. The
blockade of Pampeluna was raised on the 27th: General Hill, who
was in command at that siege, could not carry off his wounded, and
was obliged to burn part of his baggage. Twelve siege-guns (24-
pounders) were captured there. Send this to Prague, Leipzig, and
Frankfort[897].’
This ‘intelligent anticipation of the future,’ for utilization in the
armistice-negotiations going on with Austria, could not have been
bettered. Unfortunately there arrived next day another semaphore
message from Soult of the night of the 26th-27th. The Emperor has to
warn Caulaincourt that yesterday’s propaganda will not stand
criticism. ‘I have just got another “telegraphic dispatch” sent on by the
Empress from Mayence, giving another communication from Soult,
written 24 hours after the last, in which he said he would be at
Pampeluna on the 27th. The enemy lost many men and seven guns.
But nothing decisive seems to have happened. I am impatient for
more news, in order to be able to understand in detail Soult’s
dispositions, and to form from them a general idea of the
situation[898].’
Alas for human ingenuity! Soult’s next dispatch, of July 29, was
not to be of the sort that craved for publicity in the Moniteur, even with
the most judicious editing.
But to return from Dresden to Biscay, and from the head-quarters
of the Emperor to those of the ‘Sepoy General’ whom he had at last
begun to recognize as capable of ‘des projets très sensés.’
If only Wellington had been at his head-quarters at Lesaca at 11
o’clock on the morning of July 25th, and if William Stewart had been
on the spot at Maya, and had sent early news of D’Erlon’s attack,
many things might have happened differently. Wellington would have
had a long afternoon before him to concert operations, and would
have possessed information to guide him in drawing up his scheme.
Unfortunately he was absent—as we have seen—and only received
at 6 o’clock a second-hand report from Lord Dalhousie at Echalar, to
the effect that fighting was going on at Maya, with the unfortunate
addition that D’Erlon had been repulsed—a most inaccurate summary
of what had happened. Later on in the evening, not before 10 p.m.,
came Cole’s first dispatch from Roncesvalles, to say that he and
Byng were heavily engaged at 1 o’clock with a large French force,
and were holding their own. On these scanty data Wellington felt that
no conclusions could be drawn—he wrote to Graham that there must
be a great mass of French troops not yet discovered, which would
come into action on some other point on the 26th, and that his policy
would depend on where that force appeared—he could only account
for 30,000 of Soult’s men so far. He did not commit himself to any
definite guess as to the undiscovered part of the Marshal’s plan, but
from his other correspondence it is clear that he suspected an
attempt to relieve St. Sebastian by an attack on the lower Bidassoa—
a very possible solution of the problem, but not the correct one[899].
Awaiting further developments, Wellington issued no more orders
on the night of the 25th, save one to the Conde de Abispal, directing
him to send one of his two infantry divisions from in front of
Pampeluna to join Picton and Cole, and to keep the fortress
blockaded by the other. The force thus taken away would be replaced
by Carlos de España’s division, which was marching up from Burgos,
and due to arrive on the 26th. In this dispatch Wellington asked the
Conde to direct Mina to send up his infantry from Saragossa, and told
him that he was intending to order to the front the British heavy
cavalry brigades, now cantoned along the Ebro. No other movements
were settled that night; but Wellington was aware that during his
absence his Quartermaster-General, George Murray, had directed
Lord Dalhousie to have the 7th Division massed at Echalar, prepared
to move at an hour’s notice, and Charles Alten at Vera to have the
Light Division got into a similar readiness. Either would be able to
march off at dawn.
Somewhere late in the night[900] Wellington received more news,
which made the situation clearer but more unsatisfactory. The true
story of the Maya fighting came in from two sources: Hill sent a
dispatch dated from Elizondo at some hour after 6 p.m., to say that
on getting back from the Alduides he had found Stewart unable to
hold the Pass, and had bidden him to retire. Stewart, who was
wounded and unable to write, sent a verbal message, which came in
about the same time, reporting that Hill had directed him to fall back
on Elizondo and Berueta. The officers who brought this information
stated that the French were in great force, and that the 2nd Division
had been much cut up. No more reports arrived from Cole, so that the
result of the Roncesvalles fighting remained unknown.
After what must have been a very short and disturbed night’s rest,
Wellington was in the saddle by 4 a.m. on the 26th, and preparing to
ride up the Bastan to visit Hill, and to ascertain the exact measure of
the mishap at Maya. Before departing from Lesaca he gave his first
definite orders in view of the events of the previous day[901]. Maya
being lost, the 7th Division must fall back from Echalar to Sumbilla,
on the road to Santesteban: the Light Division must retire from Vera
to the west bank of the Bidassoa, and be ready to march either
towards Yanzi or towards Santesteban, as might be necessary.
Longa’s Cantabrians were to block the hill road from the Bidassoa to
Oyarzun. Graham was told to hurry on the embarkation of the siege-
train from St. Sebastian. Hill was to hold on as long as he could to the
position at Irurita, in order to keep touch with the 6th Division, which
was directed to feel towards him, and to be ready to join him if
necessary. It was to push two of its three brigades to Legasa, on the
road from Santesteban along the upper Bidassoa, which would bring
them within eight miles of Hill’s proposed line of defence at Irurita.
The third brigade of the 6th Division was to stand fast at Santesteban,
where it would be in touch with Dalhousie, when the latter should
have reached Sumbilla.
All these orders, as is obvious, are concerned only with the
measures necessary to stop D’Erlon’s advance. None of them have
any reference to the action of the other French force at Roncesvalles.
Till news should come up from Cole and Picton, it was impossible to
realize what was going on at that front, or whether the enemy was
making his main attack in that direction. There might be still (as
Wellington had guessed three days back) a violent demonstration
towards Pampeluna, intended to distract a real attempt to relieve St.
Sebastian.
And this state of ignorance with regard to the southern theatre of
operations was destined to last till late in the afternoon. Either Cole
and Picton themselves, or the officers to whom they entrusted their
dispatches, were sadly lacking in a sense of the value of the prompt
delivery of news. Wellington rode along the Bidassoa for many a mile,
till he came on Hill still holding the position of Irurita, and entirely
unmolested by the French. There were now in line the sadly reduced
remnant of the British brigades which had fought at Maya, and da
Costa’s and Ashworth’s Portuguese, with the three 7th-Division
battalions which had saved Stewart from disaster. The total made up
about 9,000 bayonets. Hill estimated[902] D’Erlon’s force at 14,000
men—a miscalculation, for even after the losses at Maya there were
still 18,000 French in line. The immediate result of the error, however,
was beneficial rather than otherwise, for Wellington considered that
Hill was in no particular danger, and let him stand, while he himself
rode southward towards the lofty Col de Velate, to seek for
intelligence from the Pampeluna front in person, since his lieutenants
had vouchsafed him none. He reached Almandoz, near the crest of
the Pass, in the afternoon, and resolved to establish his head-
quarters there for the night, as it was conveniently central between
the two halves of his army.
Soon after his arrival Wellington, being much vexed at receiving
no news whatever from the south, resolved to send the 6th Division
toward Pampeluna by the Col de Velate as a matter of precaution—
they were to march to Olague in the valley of the Lanz. The 7th
Division was to close in, to take up the ground where the 6th had
been placed, and cover Hill’s left flank[903]. That haste in these
movements was not considered a primary necessity, is shown by the
fact that Pack and Dalhousie were told that they need not march till
the morning of the 27th. For the enemy’s surprising quiescence at the
head of the Maya pass had reassured Wellington as to any danger on
this side. If D’Erlon, indeed, possessed no more than 14,000 men,
Hill with the aid of the 7th Division could easily take care of him. And
the Light Division might still be left near Lesaca, as a reserve for
Graham in case any new mass of French troops should take the
offensive on the Bidassoa.
D’Erlon’s conduct on the morning of the 26th was explicable to
himself, though inexplicable to his enemy. He had been engaged in a
most bitter fight, in which he had lost 2,000 men and more. Two
British divisions, so he wrote to Soult, were in front of him—the 2nd
and the 7th. For he had taken Barnes’s brigade for the whole of
Dalhousie’s unit—the effect of its desperate charge almost justified
him in the hypothesis. These troops had been forced to a strategic
retreat, but by no means put out of action. They must have been
joined, ere now, by the Portuguese column which Darmagnac had
sighted on its approach to Ariscun. But there were also troops on his
right, of whom he must beware: he knew that Vera and Echalar had
been held in strength, and Graham might send reinforcements in that
direction, and assemble a heavy force on his flank. Hence he
resolved to discover how matters lay by reconnaissances, before
committing himself to the march down into the Bastan and then up
the Col de Velate which his orders prescribed.
‘In my position on the pass of Maya,’ he wrote, ‘I had on my right
all the forces which the enemy had in line as far as St. Sebastian. I
had to be prudent, in order not to expose myself to a check in the
Bastan, in which the enemy was holding the strongest position. I
therefore determined to leave Abbé and Maransin in the pass, with
orders to send out reconnaissances towards Santesteban, Echalar,
and Mount Atchiola. They would profit by the halt to distribute the
half-ration of food which had just come up from Ainhoue. I sent
Darmagnac down the road to Ariscun, with orders to push a vanguard
to Elizondo, and to explore towards the passes of Ispegui and
Berderis, to see if there were any hostile force still on my left.’
An advance of six miles to Elizondo, and that by a mere advanced
guard, was all the movement that D’Erlon made this day. It was not till
the afternoon that he learnt, by Abbé’s reconnaissances, that there
were still allied troops on his right—apparently the Light Division
opposite Vera, and the 7th at Sumbilla—while Darmagnac reported
that the eastern passes were clear, but that Hill was lying across the
road beyond Elizondo in great strength. In the evening D’Erlon heard
that Soult had forced the pass of Roncesvalles, and was about to
advance: this success, he deduced, would make the enemy in front of
him give way, in fear that his positions might be taken from behind.
So he thought himself justified in ordering a general advance for the
morning of the 27th—though Maransin was still to remain for a day at
Maya, lest any allied force might move up from the west against the
pass. Thus it came that for the whole of July 26th Hill was
unmolested, and Soult’s plan for a rapid concentration round
Pampeluna became almost impossible to carry out. A whole day had
been wasted by D’Erlon, though he was not without his extenuating
circumstances.
Wellington meanwhile received at Almandoz, probably at about 8
p.m., the long-expected news from the South. They were, as we
know, most unsatisfactory: Cole reported from Linzoain, on the
Roncesvalles-Pampeluna road, that he had been driven out of the
pass by an army of 35,000 men or more, that he had not yet been
joined by the 3rd Division, and was still retreating towards Zubiri,
where he understood that Picton would meet him and take over the
command. His view of the situation was shown by a remark that if he
had not been superseded, and had been compelled to retreat past
Pampeluna, he supposed that the road towards Vittoria would have
been the right one to take[904]. This most exasperating dispatch only
reached Wellington that night by mere chance. The officer bearing it
was going to Lesaca, having no knowledge that Army Head-Quarters
had left that place: at Lanz he happened to meet the cavalry brigadier
Long, whose squadrons were keeping up the line of communication
between the two halves of the Army. Hearing from the aide-de-camp
of the sort of news that the letter contained, Long opened it and made
a copy of it, which he sent to Sir Rowland Hill, before permitting the
bearer to go on. Hill received the transcript at Berueta at 6 p.m., and
very wisely forwarded it to Wellington at Almandoz. The original was
carried on by Cole’s messenger to Santesteban, and did not reach
Wellington that night.
Thanks to Long’s and Hill’s intelligent action, the Commander-in-
Chief could grasp the whole unpleasant situation at 8 p.m. on the
26th. He sent orders to Picton at once, telling him that the enemy
must at all costs be detained: that considering the force at his
disposal, he ought to be able to check Soult for some time in front of
Zubiri: that he would be joined at once by one of O’Donnell’s divisions
from the Pampeluna blockading force, and shortly by reinforcements
coming from the Bastan (the 6th Division). Wellington himself was
intending to ride over to the right wing by the next afternoon. Till he
should arrive, Picton must send reports every few hours[905].
Unfortunately, Cole and Picton had got things into an even worse
state than could have been expected. Just as Wellington was drafting
these orders for an obstinate rearguard action, they were at 8.30 p.m.
preparing to evacuate the Zubiri position, and setting out on a night
march for Pampeluna[906].
To explain this move we must go back to the state of affairs at
Roncesvalles on the very foggy morning of July 26th. Cole, Byng, and
Morillo had abandoned, as we have already seen[907], their position on
Altobiscar and the Linduz under cover of the night, and had all fallen
into the Pampeluna road, Ross’s brigade descending from the
heights by the Mendichuri pass, the other three brigades and Morillo
moving by the chaussée past the Abbey and Burguete. Anson’s
brigade formed the rearguard, not having been engaged on the
previous day. Morillo’s outlying battalion at the Foundry of Orbaiceta
safely joined in by a hill path. Campbell’s Portuguese retired by the
way that they had come, along the Path of Atalosti, but instead of
returning to the Alduides followed a mule track to Eugui in the upper
valley of the Arga.
Cole’s long column, after completing its night march, took a much-
needed rest for many hours along the high road near Viscarret. It saw
nothing of the French till the early afternoon, when an exploring party
of chasseurs ran into the rearguard of Anson’s brigade.
What had Soult been doing between early dawn, when his
outposts ascertained that there was no enemy in front of them, and
three o’clock in the afternoon, when his cavalry rediscovered Cole?
To our surprise we find that he had been attempting to repeat his
error of the preceding day—that of sending a whole army corps along
a rugged mule track, similar to the one on which Reille’s column had
been blocked by Ross’s brigade. His original order on the 25th had
been that Reille, after seizing the Linduz, should turn along the ‘crest
of the mountains’, occupy the Atalosti defile, and push ever westward
till he could threaten the Col de Velate, the main line of
communication between the two sections of Wellington’s army. One
would have supposed that the events of the 25th on the Linduz,
where one British brigade had checked for a whole day Reille’s
column of 17,000 men in Indian file, would have taught him the
impracticability of such plans. But (as Soult’s malevolent critic, quoted
already above, observed) when the Marshal had once got his plan
drawn up on paper it was like the laws of the Medes and Persians,
and must not be altered[908].
While Clausel was directed to use the chaussée and pursue Cole
along the Pampeluna road past Roncesvalles, Burguete, and Espinal,
Reille was once more ordered[909] ‘to follow the crest of the mountains
to the right, and to try to take in the rear the hostile corps which has
been holding the pass of Maya against Count D’Erlon.’ The itinerary
seems insane: there was a mule track and no more, and Soult
proposed to engage upon it a column of 17,000 men, with a front of
one file and a depth of at least six miles, allowing for the battalion-
and brigade-intervals. The crest was not a flat plateau, but an
interminable series of ups and downs, often steep and stony,
occasionally wooded. Campbell’s brigade had traversed part of it on
the 25th, but to move a brigade on a fine day is a different thing from
moving an army corps in a fog.
Reille obeyed orders, though the fog was lying as densely upon
the mountains as on the preceding night. Apparently Soult had
supposed that it would lift at dawn—but it did not till midday.
Lamartinière’s division was left to guard the Linduz and the debouch
of the Atalosti path: Foy’s, followed by Maucune’s, tried to keep to the
crest, with the most absurd results. It was supposed to be guided by
French-Basque peasants (smugglers, no doubt) who were reputed to
know the ground. After going no more than a mile or two in the fog,
the guides, at a confusion of tracks in the middle of a wood, came to
a standstill, and talked volubly to Foy in unintelligible Basque.
Whether they had lost their way, or were giving advice, the General
could not quite discover. In despair he allowed the leading battalion to
take the most obvious track. They had got completely off the Atalosti
path, and after two miles of downhill marching found themselves on
the chaussée not far from Espinal, with the rear of Clausel’s corps
defiling past them[910]. It would still have been possible to stop the
column, for only one brigade had reached the foot of the mountain,
and Maucune and Lamartinière were still on the crest. But Reille took
upon himself the responsibility of overriding his commander’s
impracticable directions, and ordered Foy to go on, and the rest to
follow, and to fall in to the rear of Clausel’s impedimenta. ‘Il est fort
dangereux dans les hautes montagnes de s’engager sans guides et
en brouillard,’ as he very truly observed. Justifying himself in a letter
of that night to Soult, he wrote that if it were absolutely necessary to
get on the crest-path again, it could be done by turning up the Arga
valley at Zubiri, and following it to Eugui, from which there were
tracks both to the Col de Velate and to Irurita.[911]
Thus ended Soult’s impracticable scheme for seizing the Col de
Velate by marching three divisions along a precipitous mule track.
Even if there had been no fog, it is hard to believe that anything could
have come of it, as Campbell’s Portuguese would have been found at
Eugui well on in the day, and after Reille’s column would have been
much fatigued. Any show of resistance, even by one brigade, would
have checked Foy, and compelled Reille to deploy—an interminable
affair, as the fight on the Linduz upon the preceding afternoon had
sufficiently demonstrated. But to try this manœuvre in a dense fog
was insane, and Reille was quite right to throw it up.
The whole interest, therefore, of the French operation on July 26th
turns on the doing of Clausel’s column. It advanced very cautiously
down the slopes to the Abbey of Roncesvalles, discovering no trace
of the enemy save a few abandoned wounded. Having reached the
upland valley of Burguete, Clausel sent out cavalry patrols, and
found, after much searching in the fog, that Cole had gone off with his
whole force towards Espinal. His rearguard was discovered
bivouacking along the road beyond that village. When it sighted the
French it retired towards Viscarret. Clausel then ordered his infantry
to pursue, but they were far to the rear and it was only about 3 p.m.
when Taupin’s division came into touch with the light companies of
Anson’s brigade, just as they were falling back on the whole 4th
Division, drawn up in a favourable position on heights behind the Erro
river, near the village of Linzoain. The day had at last become clear
and fine. The 31st Léger, leading the French column, exchanged a
lively fusillade with the light companies, while a squadron of
chasseurs tried a charge on their flank. But both were driven off, and
Clausel halted when he saw Cole waiting for him in order of battle. It
was not till he had brought up and deployed two divisions that he
ventured to press the Allied front, and nothing serious happened till
after 4 o’clock.
Meanwhile Picton had come up from the rear, and joined Cole at
Linzoain: the head of his troops had reached Zubiri only three miles
behind. The arrival of the truculent general, looking even more
eccentric than usual, for he was wearing a tall round civilian hat
above a blue undress frock-coat, and was using a furled umbrella by
way of riding whip, was taken by the 4th-Division soldiers as a sign of
battle[912]. ‘Here comes old Tommy: now, boys, make up your minds
for a fight’ passed down the ranks[913]. But, oddly enough, this was
about the only day in Sir Thomas’s military career when he did not
take a fair risk. He certainly came up in a bellicose mood, for he
ordered Ross’s brigade to be ready to move forward when the 3rd
Division should have come up to support it. But after riding to the
front, and holding a long talk with Cole, he agreed with the latter that
it would be dangerous to fight on ground which could be turned on
both flanks, with an enemy who was known to have 35,000 men in
hand. Only part of the French were up—Reille’s divisions after their
stroll in the fog were far to the rear behind Clausel—so it would be
possible to hold on till night, and slip away in the dark. Picton wrote to
Wellington to report his decision, and does not seem in his dispatch
to have realized in the least that he was contravening the whole spirit
of his commander’s instructions of July 23rd with reference to the
‘stopping of the enemy’s progress towards Pampeluna in the event of
the passes being given up[914].’ He merely stated that he had received
these instructions too late to make it possible for him to reach
Roncesvalles, or to join Cole before the latter had evacuated his
positions[915]. As there was no favourable ground between the Erro
river and the immediate vicinity of Pampeluna, on which a smaller
force could make an effectual stand against a much larger one, he
had determined to retire at once, and proposed to ‘take up a position
at as short a distance as practicable from Pampeluna’—by which he
meant the heights of San Cristobal, only two or three miles out from
that fortress. He was thus intending to give up without further fighting
ten miles of most difficult hilly country, where the enemy could be
checked for a time at every successive ridge—though, no doubt, all
the positions could be turned one after the other by long flank
détours. But the net result was that Picton gave Soult a clear road on
the 27th, and allowed him to arrive in front of Pampeluna on that day,
whereas the least show of resistance between Zubiri and the
debouch into the plains at Huarte, would have forced the French to
deploy and waste time, and they could not have reached the open
country till the 28th. This is sufficiently proved by the extreme difficulty
which Soult found in conducting his march, even when he was not
opposed.
So determined was Picton not to fight on the Erro river, or on the
Arga, that he did not bring up his own division from Zubiri, but let it
stand, only three miles behind the line on which Cole kept up a mild
detaining action during the late afternoon hours of the 26th. Soult
attacked with great caution, and more by way of flank movements
than by frontal pressure. By evening Cole had drawn back one mile,
and had 168 casualties, all but four of them in Anson’s brigade[916].
Those of the French can hardly have been more numerous: they
seem all to have been in Taupin’s division[917].
On the afternoon of the 26th Picton had nearly 19,000 men at his
disposition[918], Soult had somewhat less, since Reille’s column was
so far to the rear that it could not get up before dark. There was no
wonder, therefore, that the enemy made no resolute attack; and it can
only be said that the Marshal was acting very wisely, for the French
force on the ground was not sufficient to move the opposing body,
until Reille should have come up; and Cole and Picton had resolved
not to give way before dark. But when the fires of the French, shining
for many miles on each side of the road, showed that they had settled
down for the night, Cole drew off his division, and retired on Zubiri,
where he passed through Picton’s troops, who were to take over the
rearguard duty, as they were fresh and well rested. Campbell’s
Portuguese dropped into the line of march from Eugui, by orders
issued to them that afternoon, and by 11 p.m. the whole corps was in
march for Pampeluna. Its departure had passed wholly unnoticed by
the enemy. Meanwhile, Wellington’s aide-de-camp, riding through the
night from Almandoz, with orders to Picton to maintain the ground
which he was abandoning, can only have met the column when it was
drawing near its destination.
It was quite early in the morning, though the sun was well up,
when the head of the retreating column reached the village of
Zabaldica, where the valley of the Arga begins to open out into the
plain of Pampeluna, between the last flanking heights which constrict
it. In front was the very ill-chosen position which Picton intended to
hold, along a line of hills which are quite separate from the main block
of the mountains, and stretch isolated in the lowland for some five
miles north-west and south-east. These are the hill of Huarte on the
right, parted from the mountains by the valley of the Egues river; the
hill of San Miguel in the centre, on the other side of the high road and
of the Arga river, and on the left the very long ridge of San Cristobal,
separated from San Miguel by the Ulzama river, which flows all along
its front.
Now these hills are strong posts in themselves, each with a good
glacis of slope in its front; the gaps between them are stopped by the
large villages of Villaba and Huarte, both susceptible of obstinate
defence; and the two flanking hills are covered in front by river-beds
—though fordable ones. But they are far too close to Pampeluna,
which is but one single mile from San Cristobal: the guns of the
fortress actually commanded at a range of only 1,200 yards, the sole
road of communication along the rear of the position. Cassan’s
garrison of 3,000 men was not large enough to furnish men for any
large sortie—though he made a vigorous sally against O’Donnell’s
blockading division on the 27th, and destroyed some of its
trenches[919]. But no army should fight with a hostile fortress less than
two miles in its rear, and commanding its line of retreat: it is surprising
that such an old soldier as Picton chose this ground—presumably he
was seduced by the fine position for both infantry and guns which it
shows looking towards the enemy’s road of arrival.
Apparently Cole had a better eye for ground than Picton, for as
they were riding together between Zabaldica and Huarte, he pointed
out to his senior the advantage that would be gained by throwing
forward the left wing of the army to a position much more remote from
Pampeluna, the hill of Oricain or Sorauren, which faces the San
Cristobal ridge from the other side of the Ulzama river[920]. This height
is the last roll of the mountains, but almost separated from their main
massif: it is only joined to the next summit by a high col at its right
centre. For the rest of its length it is separated from its neighbour-
height by a well-marked ravine. Its flanks are guarded by the beds of
the Arga to the right and the Ulzama to the left. It is well under two
miles long, about 1,000 feet high, and except at the Col has a very
formidable front of steep slopes, covered with gorse and scattered
bushes. The whole formed a strong and self-contained position,
whose weak point was that it was rather too much in advance of the
Huarte-San Miguel heights, which trend away southward, so that
when the army was drawn out its right was much ‘refused,’ and its left
very much thrown forward. It was also inconvenient that the access to
the crest from the rear was bad, a steep climb by sheep-tracks from
Oricain or Arre, up which all food or munitions would have to be
brought. From the north there was a slightly better path to the summit
from Sorauren, leading up to the small pilgrimage chapel of San
Salvador on the left end of the crest. But this would be of more use to
the assailants than to the defenders of the heights. Between the Col
and the river Arga, and close above the village of Zabaldica, there
was a spur or under-feature of the main position, which formed a sort
of outwork or flank protection to it. At the moment when the retreating
army was passing on towards Huarte, this spur was being held by
two Spanish battalions, part of the division which O’Donnell, by
Wellington’s orders, had detached to reinforce Picton. It was perhaps
the sight of this small force in a very good position which suggested
to Cole that the right policy was to prolong his line in continuation of
it, across the Col and as far as the chapel above Sorauren.
Having allowed Cole to take up his new advanced position, Picton
drew out the 3rd Division on the hill to the right of Huarte, with its
flank eastwards covered by four brigades of cavalry, which had come
up by Wellington’s orders from their cantonments on the Ebro[921].
Morillo’s Spaniards continued the line westward along the Cerro de
San Miguel, as far as Villaba: from thence the San Cristobal ridge
was occupied by the greater part of the division which O’Donnell had
drawn from the blockading lines—all, in fact, save the two battalions
in advance on the hill by Zabaldica. Later in the day two battalions
more were added from the besieging force, for Carlos de España’s
division from Castile had arrived, and relieved part of the troops
which had hitherto been observing Pampeluna. Byng’s brigade was
told off to support the 4th Division, and took post on the rear of the
summit of the Oricain hill, half a mile behind Cole. The actual fighting
line on the left was composed of Anson’s brigade on the Col, next to
the Spaniards on the lower spur, of Campbell’s Portuguese upon the
central stretch of the heights (except one battalion which was sent to
support the Spaniards below)[922], and of Ross’s brigade holding the
left. Stubbs’s Portuguese were in rear of Campbell’s, except the 7th
Caçadores, which was detached to the front and held the ground
about the chapel of San Salvador. The divisional battery (Sympher’s
of the K.G.L.) was placed far down the right side of the hill, below and
behind Byng’s brigade, in a position from which it could sweep the
high road from Zabaldica to Arleta. Cole’s tactical dispositions were in
the complete Wellingtonian style, with the light companies and
caçadores thrown out some way down the slope, far in advance of
the main force, whose battalions were drawn back well behind the
sky-line, so as to be invisible till the last moment to enemies storming
the hill. Soult followed up the retreating Allies at such a slow pace
that the whole of Picton’s troops were settling into their ground before
the enemy came in sight[923].
The slow advance of the French was due to the accumulation of
such a large force in a narrow valley provided with only one road. The
Marshal made an attempt to relieve the congestion, by ordering that
the chaussée should be left to Clausel and to the cavalry and
impedimenta in his rear, while Reille’s divisions should move on the
east bank of the Arga by local paths between the villages. The
excellent intention of securing room for both columns to move freely
had no good result. Clausel arrived in front of Zabaldica by 9 a.m.[924]
But Reille was nowhere in sight. His report fully explains his absence:
he had obeyed orders by turning up into the hills a mile and a half
beyond the village of Erro. ‘This direction rendered the march of the
three divisions extremely slow and difficult. They found no road, and
had to tramp through brushwood, climb steep slopes, or to follow
tracks obliterated by recent rain. At last Count Reille took the decision
to abandon the high ground. The 1st Division (Foy) dropped down to
the village of Alzuza on the extreme left. The 7th Division (Maucune)
re-descended into the valley of the Arga, a little above Iroz, where it
bivouacked. The 9th Division (Lamartinière) also came down into the
valley opposite Larrasoana, and kept along the high road to Iroz,’
where it fell in with the rear of Clausel’s column late in the day. The
only result of Soult’s precaution had been to put Reille out of the
game on the 27th, just as on the 26th.
For the whole of the morning hours, therefore, Soult had only
Clausel’s corps at his disposition, a fact which accounts for the
unenterprising character of his action. But that the 27th was a very
slack day on the French side was not Clausel’s fault. On arriving at
Zabaldica, and discovering that the heights of Oricain were held in
strength, he did not wait for the Marshal’s orders, but began to form a
line of battle parallel to Cole’s front, along the mountain opposite.
Halting Conroux’s division on the high road in face of the hill held by
the Spaniards, he pushed Taupin’s and Vandermaesen’s divisions up
the slopes, with cavalry detachments feeling the way in front of them,
till they had lined the whole ridge, and their right was overlooking
Sorauren and the valley of the Ulzama. He then sent down to ask the
Marshal’s leave to attack, saying that he could see from the summit
behind his line large baggage trains moving away along the Vittoria
road in the plain of Pampeluna, and bodies of troops in motion
northward[925]—the enemy was about to raise the siege, and was only
offering a rearguard action in order to cover the retreat of his
impedimenta. If pressed he would give way at once[926].
Soult did not believe this, and very rightly; but being pressed by
repeated messages he mounted up to the heights behind Clausel’s
front at 11 a.m.: if he had chanced to notice it, he was just in time to
see a solitary horseman ride up the north-western slope of the hill of
Oricain, and to hear the whole of the Allied troops aligned opposite
him burst out into a storm of tempestuous cheering. Wellington had
come upon the ground. Soult heard the noise, but (as his dispatches
show) did not guess its precise cause. He thought that
reinforcements had just come up for Cole.
The story of Wellington’s eventful ride from Almandoz to Sorauren
is a very interesting one. Much irritated at receiving no further news
from Picton, he had mounted at sunrise and ridden over the Col de
Velate, taking with him only George Murray, his Quartermaster-
General, his Military Secretary Fitzroy Somerset, and three or four
other officers: the bulk of the head-quarters staff was to follow at
leisure. On arriving at Lanz, the first village on the south side of the
pass, they heard rumours of Picton’s continued retreat, though they
do not seem to have met the aide-de-camp whom he had sent off on
the preceding night to report it. This news was so unexpected and
vexatious that Wellington halted for a moment, to send back orders to
Hill to the effect that it was conceivable that affairs might go badly on
the Pampeluna front. If so, the whole right or southern wing of the
army might have to swing back to the line Yrurzun-Tolosa, and Hill
would have to direct his own two divisions, and also Dalhousie, and
Pack, with all the artillery and baggage, to fall back westward on
Lizaso and Lecumberri, instead of coming over the Col de Velate
towards Pampeluna. The Light Division, too, might have to leave the
neighbourhood of the Bastan, and to retire to Zubieta on the
Oyarzun-Lecumberri road, in order to keep up the touch between the
main army and Graham’s force in front of St. Sebastian. The latter
general, however, was not to move, unless matters went very badly
indeed, as the blockade of St. Sebastian must be kept up till the last
possible minute. But previous orders were to stand, unless and until
the Commander-in-Chief should send new ones: in particular Pack
and the 6th Division were expected at Olague, and the batteries of

Artificial Intelligence in The Gulf Challenges and Opportunities Elie Azar Full Chapter PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Artificial Intelligence in The Gulf Challenges and Opportunities Elie Azar Full Chapter PDF

Uploaded by

Copyright:

Available Formats

Artificial Intelligence in the Gulf:

Challenges and Opportunities Elie Azar

The Ethics of Artificial Intelligence_ Principles,

Oil and Oilseed Processing: Opportunities and

International Management: Strategic Opportunities and

Artificial Intelligence and Data Science in

Corporate Sustainability in Africa: Responsible

Artificial Intelligence and Intellectual Property Reto

Precision Health and Artificial Intelligence Arjun

Blockchain in the Global South: Opportunities and

ISBN 978-981-16-0770-7 ISBN 978-981-16-0771-4 (eBook)

© Gulf Research Centre Cambridge 2021

Cover illustration: @Buena Vista Images/GettyImages

This volume is based on a selection of papers presented at the “Artificial

United Arab Emirates Elie Azar

Part II Data, Governance and Regulations

Part III Existing Opportunities and Sectoral

Part IV Society, Utopia and Dystopia

Helen Abadzi University of Texas, Arlington, TX, USA

Toufic Mezher Khalifa University of Science and Technology, Abu

Fig. 4 Mr Usta heatmap overlay on the city of Dubai, showing

An Introduction to AI in the GCC

Elie Azar and Anthony N. Haddad

© The Author(s), under exclusive license to Springer Nature 3

As a consequence, these countries have taken a progressive stance

While headline-grabbing in nature, these initiatives are not isolated,

1.3 Objectives, Scope, and Target Audience

schools of thought on AI, and the power of AI to help deliver well-being

Framework of Study and Book Organization

Elie Azar and Anthony N. Haddad

© The Author(s), under exclusive license to Springer Nature 9

VI: Conclusion—which presents an overview of the topics covered in

on the healthcare sector of GCC states and the emerging concept of

“Fatwas” are studied, which are legal opinions expressed by a Muslim

Data, Governance and Regulations

Public Sector Data for Academic Research:

Aleksandar Abu Samra, Toufic Mezher, and Elie Azar

A. Abu Samra · T. Mezher · E. Azar (B)

© The Author(s), under exclusive license to Springer Nature 15

challenges (Harvey & Tulloch, 2006). The problems of hardly approach-

The main goal of this chapter is to comprehensively evaluate and

– Understand what logistical, communicational, legal, and personal

Sharing knowledge between individuals and organizations is a complex

2.2 Open Data and Open Government Data (OGD)

2.3 Information Sharing Trends in UAE

Table 1 Resources of best practices in OGD

Source Type Reference

International Organizations Open Government Data Toolkit (World Bank, 2016)

framework is then presented to guide efforts toward implementing such

3.1 Interview Design

1. Getting to Know the Interviewee—To briefly give a context of the

Fig. 1 Methodology development

The interview with public officials had the objectives of understanding

1. Getting to Know the Interviewee—A brief introduction to the

2. Data Collection Processes —Understanding what type of data the

After developing the first versions of both interviews, test-runs were

3.2 Data Collection and Scope of Work

people in other public authorities. The investigator would then email or

3.3 Data Analysis and Identification of Barriers

3.4 Understand Challenges to Implementing Best Practices

3.5 Propose Solution Framework

1. Strategy—Describing the proposed policy and actions performed by

To validate such a solution framework, the author held meetings with