"Generative AI is changing the world we live in. It represents the most disruptive and transformative technology of our lifetime. It will revolutionize how we interact with technology and how we work."
That quote was from Bill Wong, research fellow at Info-Tech Research Group. He and his team recently found that there is a rapidly growing responsibility of organizations to safeguard users against potential risks associated with AI; this includes misinformation, unfair bias, malicious uses and cybersecurity threats. However, many existing risk and governance programs within organizations have not been designed to anticipate the introduction of AI applications and their subsequent impact.
“Generative AI has demonstrated the ease of creating misinformation and deepfakes, and it can be misused to threaten the integrity of elections,” said Wong.
Info-Tech also found that establishing and operationalizing responsible AI principles to govern AI development and deployment will be crucial for organizations. This involves creating a framework that includes ethical guidelines, transparency, accountability, and fairness in AI applications.
“Effective regulation of AI on a global scale is likely to necessitate international cooperation across governments and regions,” Wong added.
So, this prompted Wong and the research team at Info-Tech to come up with a new blueprint, “Prepare for AI Regulation.” The resource addresses the urgent need for organizations to stay ahead of impending regulations.
Info-Tech outlined six guiding principles and corresponding actions that IT leaders can follow to mitigate risks and ensure compliance with emerging AI regulations.
Data Privacy is important in AI development. Organizations must conduct thorough assessments to understand applicable privacy laws and create detailed data flow maps to track personal data. Minimizing data collection and storage is crucial to protect sensitive information.
Fairness and Bias Detection is another critical aspect. Identifying potential sources of bias in data and algorithms is essential. Organizations must also comply with accessibility and inclusivity laws and ensure diversity in training data to promote fairness and avoid discrimination.
Explainability and Transparency are key to building trust in AI systems. Organizations should design AI systems that provide clear explanations of decision-making processes. Disclosing training data and methodologies, as well as enforcing data labeling practices, can enhance transparency.
Now let’s discuss Safety and Security. Adopting responsible design, development and deployment practices is essential to prevent risks. Providing clear information to deployers and promoting cybersecurity measures can help safeguard AI systems.
As for Validity and Reliability, they are crucial for ensuring AI systems perform as intended. Continuous monitoring, evaluation and validation are necessary to maintain accuracy. Provenance tracking and quality assessments of training and collected data can help ensure reliability.
Lastly, there is Accountability, which is essential for responsible AI. Implementing human oversight and review, assigning risk management responsibilities and integrating AI governance with the overall risk management framework can help organizations maintain accountability and ethical AI practices.
Info-Tech's blueprint provides practical guidance for organizations looking to implement AI responsibly. It outlines strategies to ensure regulatory compliance and ethical AI usage. It also helps organizations get the most out of AI’s potential while mitigating risks.
Edited by
Alex Passett
