Ethical Risks in generative AI - Mitigation Plans
Praveen kumar
Quality Assurance I CMMI I InfoSec I Process Excellence I Operation Management | Author of 'Myths of Spice Route'
As AI become part of our personal and professional life, the ethical implications of technology needs to be looked closely.
For example,
If an AI created Design gets recognized, who should be rewarded?
The developers of AI design or the users who have created the output.
If AI driven automated Vehicle should make a choice between hitting a pedestrian and breaking the traffic Rule to avoid an accident? Which one it should take. Should the algorithm allow breaking traffic Rules.
If there is an inherent bias in the AI tool, how do we know?
The above scenarios highlights the ethical Risks in AI implementation. And if we probe further, a Pandora of questions will evolve without definite answers.
Hence Identifying and mitigating ethical Risks in AI becomes paramount.
Below are some steps we can initiate towards securing AI tools.
Ethical Frame work: Organization should create an AI framework based on what they consider as Value. Values like Fairness, Transparency, accountability should be part the Frame work while creating AI tools.
Training: Ethical training should be provided to programmers to avoid bias during creation of AI tools. For example, a recruitment tools made with the help of AI should not be having an bias on selecting candidates. This can be ensured only by providing appropriate training to programmers.
Vulnerability assessment: Vulnerability assessment to be carried out before any AI tool roll out to identify Potential Risks. Scenarios related to Data security can be easily identified through Vulnerability assessment.
Monitoring: Regular Monitoring on AI tools to be done to identify any bias in the system. Also, results of monitoring can be used to improve AI tool.
As the generative AI technology evolves, the above monitoring activities also should evolve to meet the changing requirements.