top of page

"Critical Issues About A.I. Accountability Answered" CMR publishing.


I am proud of the publication of the article "Critical Issues about AI Accountability Answered" by the California Management Review. with Mostafa Sayyadi and Michael Provitera.

Firms need to face the AI system adoption; there are all the hypotheses related to the accountability for the outputs of innovative technologies, AI and machine learning.

Making AI Systems Responsible

Artificial intelligence (AI) is being used in businesses, and there are worries about what happens when AI makes mistakes or harms people. An important question is who should be held responsible when AI does bad things.

Usually, executives and managers are accountable for plans and results. But AI can act on its own without people overseeing it. If an AI chatbot gives wrong info or an AI security camera unfairly targets certain people, who should be blamed?

Ways to Make AI Accountable

Some think AI developers should be responsible. However, AI is often created by many parties over time.

Making the users liable makes sense, but they might not understand how AI decides things.

Having accountability shared by developers, users and companies is another idea. But this could water down duties if who is responsible is not clear.

Processes like testing AI, auditing algorithms, and having oversight committees can improve accountability.


  • Thoroughly testing AI before use can find biases.

  • Regularly auditing algorithms watch for new risks.

  • Oversight committees can make rules, check compliance, and step in if needed.

  • Explainable AI is key too. It shows how AI makes suggestions, building trust in opaque systems. Laws and rules for banks and other high-risk areas will also shape accountability.


What Executives and Boards Should Do

Experts say boards should require AI testing, audits, human oversight for risky AI, and checking for unfair impacts. Staying up-to-date on evolving rules is vital. In business, advisors say boards should actively monitor AI risks and have detailed oversight plans rather than handing off all control.

Executives may not grasp AI technical details fully. But they must take responsibility for any AI used. Shared accountability can spread liability, but unclear divisions weaken it. Boards and leaders should demand explainability and implement strong governance to maintain accountability.

Ideas for AI Accountability Rules?

Our proposal is based on internal steps like assessing risks before deployment, ongoing monitoring, response plans after problems, and mapping out accountability across roles. Another idea is external accountability through oversight boards, required audits, explainability standards, and community input.

Combining internal risk reduction steps and external transparency via audits and advisory boards can give comprehensive accountability. Testing, explainability and governance are key supports that allow innovation while keeping accountability.

The Way Forward

AI accountability models are still debated. Simple solutions like making only developers liable have flaws. Executives can’t avoid responsibility for AI systems, even if complex. Shared accountability, oversight processes, audits, explainability rules, clear policies and updated regulations will shape the future.

Leaders should use available tools to implement AI responsibly, even with imperfect knowledge. A proactive focus on transparency, assessing potential impacts, and strong governance will build public trust. With effort, companies can maintain accountability while benefiting from advanced AI.

Regulations are an important part of accountability, but they are not enough.

Recent Posts
No tags yet.
bottom of page