Take part in this OECD survey to gather insights from the private sector on AI adoption and AI governance. 👉 https://lnkd.in/er-Ts3-V The survey is divided into two sections, comprises 13 questions, and should take approximately 15 minutes to complete. The results will inform the OECD Global Strategy Group (GSG) Meeting 2024, which will take place on 15-16 October and the work of the OECD AI Futures Expert Group. 📅 Participants must complete the survey by 1 October 2024. Business at OECD (BIAC) Nicole Primmer Maylis Berviller Ulrik Vestergaard Knudsen Jerry Sheehan Audrey Plonk Karine Perset Luis Aranda Jamie Berryhill Lucia Russo Noah Oder John Leo Tarver ⒿⓁⓉ Rashad Abelson Angélina Gentaz Valéria Silva Bénédicte Rispal Eunseo Nikolas S. Sarah Bérubé #survey #aigovernance #industry #artificialintelligence #oecd
OECD.AI
Affaires étrangères
Paris, Île-de-France 36 379 abonnés
OECD.AI is a platform to share and shape trustworthy AI. Sign up below for email alerts and visit our blog OECD.AI/WONK.
À propos
Visit our blog, the AI Wonk: https://oecd.ai/wonk/ The OECD AI Policy Observatory is a tool at the disposal of governments and businesses that they can use to implement the first intergovernmental standard on AI: the OECD AI Principles. The OECD AI Principles focus on how governments and other actors can shape a human-centric approach to trustworthy AI. The Observatory includes a blog for its group of international AI experts (ONE AI) to discuss issues related to defining AI and how to implement the OECD Principles. OECD countries adopted the standards in May 2019, along with a range of partner economies. The OECD AI Principles provided the basis for the G20 AI Principles endorsed by Leaders in June 2019. OECD.AI combines resources from across the OECD, its partners and all stakeholder groups. OECD.AI facilitates dialogue between stakeholders while providing multidisciplinary, evidence-based policy analysis in the areas where AI has the most impact. As an inclusive platform for public policy on AI – the OECD AI Policy Observatory is oriented around three core attributes: Multidisciplinarity The Observatory works with policy communities across and beyond the OECD – from the digital economy and science and technology policy, to employment, health, consumer protection, education and transport policy – to consider the opportunities and challenges posed by current and future AI developments in a coherent, holistic manner. Evidence-based analysis The Observatory provides a centre for the collection and sharing of evidence on AI, leveraging the OECD’s reputation for measurement methodologies and evidence-based analysis. Global multi-stakeholder partnerships The Observatory engages governments and a wide spectrum of stakeholders – including partners from the technical community, the private sector, academia, civil society and other international organisations – and provides a hub for dialogue and collaboration.
- Site web
-
https://oecd.ai/
Lien externe pour OECD.AI
- Secteur
- Affaires étrangères
- Taille de l’entreprise
- 11-50 employés
- Siège social
- Paris, Île-de-France
- Type
- Administration publique
- Fondée en
- 2020
Lieux
-
Principal
2 rue André Pascal
75016 Paris, Île-de-France, FR
Employés chez OECD.AI
Nouvelles
-
Today, OECD Deputy Secretary-General Ulrik Vestergaard Knudsen spoke at the GAIN Summit in # Saudi Arabia, touching upon the OECD's role in helping countries manage the benefits and risks of AI. With the rapid development of generative AI, participants at the Global AI Summit recognise the need to balance theory and reality when discussing AI's potential. The #OECD works with partners beyond its membership to address artificial intelligence's problematic questions and help countries ensure the global interoperability of national responses. Together, we can shape #artificialintelligence into a tangible reality that benefits humanity. Visit the Summit online: https://lnkd.in/dZD5wH8X Ulrik Vestergaard Knudsen Vilas Dhar Jerry Sheehan Audrey Plonk Karine Perset Luis Aranda Jamie Berryhill Lucia Russo John Leo Tarver ⒿⓁⓉ Rashad Abelson Sara Fialho Esposito Nikolas S. Sarah Bérubé #trustworthyai #gainsummit
-
-
Public consultation on risk thresholds for advanced AI systems 📅 DEADLINE EXTENSION: 1 OCTOBER https://lnkd.in/e98Pzw-b The OECD is collaborating with diverse stakeholders to explore potential approaches, opportunities, and limitations for establishing risk thresholds for advanced AI systems. To inform this work, we are holding an open public consultation to obtain the views of all interested parties. We are interested in hearing your thoughts on the following key questions: ❓ What publications or other resources have you found helpful on AI risk thresholds? ❓ To what extent do you believe AI risk thresholds based on compute power are adequate and appropriate to mitigate risks from advanced AI systems? ❓ To what extent do you believe other AI risk thresholds would be valuable, and what are they? ❓ What strategies and approaches can governments or companies use to identify and set specific thresholds and measure real-world systems against those thresholds? What requirements should be imposed for systems that exceed any given threshold? ❓ What else should the OECD and collaborating organisations consider concerning designing and/or implementing AI risk thresholds? 📅 10 SEPTEMBER DEADLINE TO PARTICIPATE https://lnkd.in/e98Pzw-b Francesca Rossi Stuart Russell Michael Schönstein Ulrik Vestergaard Knudsen Jerry Sheehan Audrey Plonk Celine Caira Luis Aranda Jamie Berryhill Lucia Russo Noah Oder John Leo Tarver ⒿⓁⓉ Rashad Abelson Angélina Gentaz Valéria Silva Bénédicte Rispal Johannes Leon Kirnberger Eunseo Dana Choi Pablo Gomez Ayerbe Sara Fialho Esposito Nikolas S. Sarah Bérubé Guillermo H. #airisk #aisafety #trustworthyai #oecd #risk
-
-
The AI Standards Hub is enhancing the OECD Catalogue of Tools and Metrics for Trustworthy AI with AI-specific standards from its Standards Database. The Hub will continue to contribute as new standards become available. Standards can take many forms and address different parts of the AI development lifecycle, including: ✳️ foundational standards to establish key terminology; ✳️ process guidelines to describe sound, repeatable organisational practices; ✳️ technical measurement standards and the associated testing and measurement techniques to benchmark AI systems; ✳️ interfacing and networking architectures to ensure AI interoperability. Find out more in this blog post by Paul Khullar from The Alan Turing Institute. Ulrik Vestergaard Knudsen Jerry Sheehan Audrey Plonk Karine Perset Celine Caira Luis Aranda Jamie Berryhill Lucia Russo Noah Oder John Leo Tarver ⒿⓁⓉ Rashad Abelson Angélina Gentaz Valéria Silva Bénédicte Rispal Johannes Leon Kirnberger Eunseo Dana Choi Pablo Gomez Ayerbe Sara Fialho Esposito Nikolas S. Sarah Bérubé Guillermo H. #trustworthyai #alanturinginstitute #artificialintelligence #standards #aistandards
-
❗FINAL DAYS TO PARTICIPATE in the OECD’s pilot on monitoring the application of the G7 Code of Conduct for Organisations Developing Advanced AI Development❗DEADLINE 6 SEPTEMBER 📅 https://lnkd.in/eDgwNsy2. The OECD will test a reporting framework intended to gather information about how organisations developing advanced AI systems align with the Actions of the Code of Conduct. This pilot marks a significant milestone in international cooperation on AI governance, reflecting the G7’s commitment to promoting the safe, secure, and trustworthy development, deployment, and use of advanced AI systems. READ MORE IN THIS BLOG POST https://lnkd.in/eDgwNsy2 by Audrey Plonk Karine Perset and Sara Fialho Esposito André Meyer-Vitali Ulrik Vestergaard Knudsen Jerry Sheehan Celine Caira Luis Aranda Jamie Berryhill Lucia Russo Noah Oder John Leo Tarver ⒿⓁⓉ Rashad Abelson Angélina Gentaz Valéria Silva Bénédicte Rispal Johannes Leon Kirnberger Eunseo Dana Choi Pablo Gomez Ayerbe Nikolas S. Sarah Bérubé Guillermo H. #g7 #hiroshimaprocess #oecd #trustworthyai #aipolicy #aisystems
-
-
AI could transform disaster response. How do we ensure it reaches those who need it most? Blog post 👉 https://lnkd.in/eU_GJ3MB As leading economists consistently note, getting AI to serve the needs of the world's poorest and most vulnerable is anything but guaranteed. Inspired by partnerships that successfully channel the fruits of scientific progress to sectors and communities underserved by market mechanisms, the UN-led DISHA programme (Data Insights for Social and Humanitarian Action) takes a coalition-based approach to overcoming the barriers that humanitarian organisations face in the AI ecosystem. Katya Klinova from the UN Secretary-General's Innovation Lab and Yolanda Botti-Lodovico from The Patrick J. McGovern Foundation discuss the project and how public infrastructure can enable the success and scale of AI-for-good initiatives. United Nations Global Pulse United Nations Global Pulse Asia Pacific UN 2.0 | United Nations Vilas Dhar Nick Cain Ahmed Khamis El Saeed Kersten Jauer Talea von Lupin Ulrik Vestergaard Knudsen Jerry Sheehan Audrey Plonk Karine Perset Celine Caira Luis Aranda Jamie Berryhill Lucia Russo John Leo Tarver ⒿⓁⓉ Rashad Abelson Angélina Gentaz Valéria Silva Bénédicte Rispal Sara Fialho Esposito Nikolas S. Sarah Bérubé #unitednations #aiforgood #artificialintelligence #naturaldisasters #unitednations
-
-
📣 Partnership on AI is seeking feedback on its Vendor Engagement Guidance & Transparency Template AI relies heavily on enriched datasets, which rely heavily on data enrichment workers. These people work behind the scenes to label, categorize, annotate, or otherwise enrich datasets, which are then used to train AI systems. This work is crucial to the development of AI technology but mostly overlooked and undervalued. Research shows that these workers often face unfair working conditions, low wages, unclear expectations, and little support in their work. PAI outlined a five-step “Path for Developing Responsible AI Supply Chains” based on research, insights and conversations with its broader partner network. These initial steps enable better accountability and governance across the data supply chain: 1️⃣ Adopt PAI’s Data Enrichment Guidelines 2️⃣ Introduce Internal Governance 3️⃣ Promote Consistent Practices Across the Supply Chain 4️⃣ Publish Transparency Reports 5️⃣ Include Workers’ Voices when Evolving Best Practices PAI shared resources for public comment that correspond with steps 3 and 4: the Vendor Engagement Guidance to help AI-developing companies promote responsible practices with downstream actors and the Transparency Template to outline what companies should be monitoring and reporting on regarding their data enrichment practices. As the #oecd develops its Due Diligence Guidance for Responsible AI, it aims to integrate best practices from resources like this to support all companies developing and using AI systems. 📅 GIVE YOUR FEEDBACK AND REGISTER FOR THE 12 SEPTEMBER PUBLIC FEEDBACK SESSION 👇 https://lnkd.in/eVbNrBCz Nozha Boujemaa Rebecca Finlay Sebastian Hallensleben Andrea Renda Stephanie Ifayemi Sonam Jindal Ulrik Vestergaard Knudsen Jerry Sheehan Audrey Plonk Karine Perset Celine Caira Luis Aranda Jamie Berryhill Lucia Russo Noah Oder John Leo Tarver ⒿⓁⓉ Rashad Abelson Angélina Gentaz Valéria Silva Bénédicte Rispal Johannes Leon Kirnberger Eunseo Dana Choi Pablo Gomez Ayerbe Sara Fialho Esposito Nikolas S. Sarah Bérubé Guillermo H. #ai #enricheddata #duediligence #dataenrichmentworkers #partnershiponai #feedback
Protecting AI’s Essential Workers: Introducing our Vendor Engagement Guidance & Transparency Template
partnershiponai.org
-
PUBLIC CONSULTATION ON RISK ⚠️ THRESHOLDS https://lnkd.in/e98Pzw-b The OECD is joining forces with diverse stakeholders to explore potential approaches, opportunities and limitations for establishing risk thresholds for advanced AI systems. To inform this work, we are holding an open public consultation to obtain the views of all interested parties. We are interested in hearing your thoughts on the following key questions: ❓ What publications or other resources have you found helpful on AI risk thresholds? ❓ To what extent do you believe AI risk thresholds based on compute power are adequate and appropriate to mitigate risks from advanced AI systems? ❓ To what extent do you believe other AI risk thresholds would be valuable, and what are they? ❓ What strategies and approaches can governments or companies use to identify and set specific thresholds and measure real-world systems against those thresholds? What requirements should be imposed for systems that exceed any given threshold? ❓ What else should the OECD and collaborating organisations keep in mind concerning designing and/or implementing AI risk thresholds? 📅 10 SEPTEMBER DEADLINE TO PARTICIPATE https://lnkd.in/e98Pzw-b Francesca Rossi Stuart Russell Michael Schönstein Ulrik Vestergaard Knudsen Jerry Sheehan Audrey Plonk Celine Caira Luis Aranda Jamie Berryhill Lucia Russo Noah Oder John Leo Tarver ⒿⓁⓉ Rashad Abelson Angélina Gentaz Valéria Silva Bénédicte Rispal Johannes Leon Kirnberger Eunseo Dana Choi Pablo Gomez Ayerbe Sara Fialho Esposito Nikolas S. Sarah Bérubé Guillermo H. #airisk #aisafety #trustworthyai #oecd #risk
Seeking your views: Public consultation on risk thresholds for advanced AI systems – Deadline 10 September
oecd.ai
-
A warm thanks to the team at #KDD 2024 European Data Science Day in Barcelona. It was an honour for Karine Perset to present our AI Incidents Monitor. The European Data Science Day offered a full day focused exclusively on innovative KDD-relevant research and development projects in Data Science, Machine Learning and Language Technologies, and partnerships in the social and physical sciences/arts, to show the state-of-the-art in research and applications. As AI-related regulation takes shape, policymakers need evidence and foresight for evidence-based decision-making. AIM tracks AI incidents mentioned in the press and allows users to filter by types of incidents, concepts and keywords. This way, anyone can see what issues are taking shape with AI as they happen and grow. While tracking incidents in the press is not a perfect solution, it is an excellent first step. In the future, we want to allow direct submissions formulated methodically so that incidents are defined and reported in the same way everywhere. VISIT AIM HERE: https://lnkd.in/eH6REF_z Special thanks to: Francesco Bonchi Ricardo Baeza-Yates Rajesh Parekh Claudia Perlich Shipeng Yu Marko Grobelnik joao Gama Fosca Giannotti Karthik Subbian Ulrik Vestergaard Knudsen Jerry Sheehan Audrey Plonk Luis Aranda John Leo Tarver ⒿⓁⓉ Bénédicte Rispal Eunseo Dana Choi Pablo Gomez Ayerbe Sara Fialho Esposito Nikolas S. Sarah Bérubé #KDD2024 #trustworthyai #aipolicy #datascience #artificialintelligence
-
-
🛠️🧰⚙️🔧 TOOLS FOR TRUSTWORTHY AI 🛠️🧰⚙️🔧 BELLS - Benchmarks for the Evaluation of LLM Safeguards - 👉 https://lnkd.in/eWcFA45r CeSIA's BELLS project's overall goal is to rigorously evaluate input-output safeguards, promoting wider adoption of oversight techniques to make LLMs more reliable and safe. BELLS evaluates safeguards that detect undesired behaviours in model inputs and outputs for large language models (LLMs). These safeguards serve as a testbed for scalable oversight for #LLMs. BELLS evaluates two types of safeguards: #jailbreak and #hallucination detectors. The evaluations assess safeguard performance along two axes: Performance on established datasets of jailbreaks and hallucinations. This tests how well safeguards detect known failure modes. Generalization, specifically for jailbreaks, to new types published after the safeguard's release. This is a proxy for the tool's ability to catch future and unknown failure types. BELLS ranks the most effective safeguards on a leaderboard to help users select the tools that offer the most robust protection against LLM harms. Thanks to Diego Dorn, Alexandre Variengien and Charbel-Raphaël Segerie. Ulrik Vestergaard Knudsen Jerry Sheehan Audrey Plonk Karine Perset Celine Caira Luis Aranda Jamie Berryhill Lucia Russo Noah Oder John Leo Tarver ⒿⓁⓉ Rashad Abelson Angélina Gentaz Valéria Silva Bénédicte Rispal Johannes Leon Kirnberger Eunseo Dana Choi Pablo Gomez Ayerbe Sara Fialho Esposito Nikolas S. Sarah Bérubé Guillermo H. #trustworthyai #artificialintelligence #oecd
BELLS - Benchmarks for the Evaluation of LLM Safeguards
oecd.ai