Friends, of late, Artificial Intelligence(AI) seems to be taking center stage in any discussion - in person or offline. There is a lot of interest and engagement from companies across the board in how AI can help them manage their projects better. From organizational leadership to project team members are investigating ways of integrating AI and Large Language Models such as ChatGPT in their day-to-day work. AI is not letting them down, and we see automation around many tasks being done and accepted by the project leads. This is a great example of how new technology starts to penetrate ways of working. That said, who should take ownership of the decisions made by or with the help of these AI applications? For example, in healthcare, decision-making can "make" or "break" life! How are we driving ethical behavior while working with these LLMs?
I view this as an onion-like approach (similar to how cyber security threats are addressed).
At the highest level, government and industry watchdogs should put guardrails (NOT prescriptive regulations) in place to help everyone understand what is clearly outside the box.
Companies should also create policies for AI the same as are done for data management and security.
Finally, teams need to be staffed with SMEs who have the experience and capabilities to vet the input data and the results.
Kiron, I cant agree more. I like your analogy of onion peel. That said, there has to be clear ownership, especially when it comes to a negative event in healthcare. Let's for a moment assume all the peels of the onion are in place and there is an adverse event in patient's treatment. Who takes the ownership?
Being a physician, Healthcare is just an example and comes naturally. I am sure there must be similar situations in other domains as well.
...
1 reply by Kiron Bondale
Oct 11, 2023 7:08 AM
Kiron Bondale
...
Deepa, I'd suggest that this would be no different than any other adverse outcome resulting from a project or operational activity. Most regulated companies have roles dedicated to such events and I'd feel that AI-related errors would fall under their jurisdiction. This would require their teams to be trained in understanding specific nuances related to AI.
As far as accountability goes, it really depends on the nature of the error. If the project deliverables were correct and it was an operational issue related to the use of those deliverables it would likely fall to the business leader responsible for that capability.
Ownership of the decisions made by AI should be owned by both the Implementers of the AI solution(as they should only implement what they completely understand, by implementing they are implicitly taking the ownership) as well as developers of the AI Solution(as there development may impact the life of a being) and definitely not the AI itself as it is just a Tool Saving Changes...
Kiron, I cant agree more. I like your analogy of onion peel. That said, there has to be clear ownership, especially when it comes to a negative event in healthcare. Let's for a moment assume all the peels of the onion are in place and there is an adverse event in patient's treatment. Who takes the ownership?
Being a physician, Healthcare is just an example and comes naturally. I am sure there must be similar situations in other domains as well.
Deepa, I'd suggest that this would be no different than any other adverse outcome resulting from a project or operational activity. Most regulated companies have roles dedicated to such events and I'd feel that AI-related errors would fall under their jurisdiction. This would require their teams to be trained in understanding specific nuances related to AI.
As far as accountability goes, it really depends on the nature of the error. If the project deliverables were correct and it was an operational issue related to the use of those deliverables it would likely fall to the business leader responsible for that capability.
Thanks and I agree with you on the ownership. I guess, even with guardrails around, things will need to evolve for AI-driven decision-making. Saving Changes...
Ming YeungAdjunct Professor & Acting COO/CPO/CRO (contract)| Blockchain Venture Capital Inc.Toronto, Ontario, Canada
Deepa, the firms should take ownership via internal policies and procedures while their respective regulatory authorities and/or self-regulatory organizations, where appropriate and applicable, would impose boundaries and uphold governance through relevant laws, guidelines, and best practices.
...
1 reply by Dr. Deepa Bhide
Oct 11, 2023 9:26 PM
Dr. Deepa Bhide
...
Ming, thank you and I am sure things will evolve as we see progress in these technologies and newer use cases.
Saving Changes...
James HallPMO/Project Management| Independent & AvailableValencia, Ca, United States
Every time I log into ChatGPT it warns me that the results of my search may be inaccurate. While we expect AI providers to clearly state how their model is built, its strengths and vulnerabilities, it is on the user to read, understand, and clean-up the AI output. There might come a time when we rely on AI "data" like we rely on a data feed about the stock market, but until then, data quality and the deliverable is all on us.
...
1 reply by Dr. Deepa Bhide
Oct 11, 2023 9:26 PM
Dr. Deepa Bhide
...
Thanks James and I agree with you. We all need to be cautious.
Every time I log into ChatGPT it warns me that the results of my search may be inaccurate. While we expect AI providers to clearly state how their model is built, its strengths and vulnerabilities, it is on the user to read, understand, and clean-up the AI output. There might come a time when we rely on AI "data" like we rely on a data feed about the stock market, but until then, data quality and the deliverable is all on us.
Thanks James and I agree with you. We all need to be cautious. Saving Changes...
Deepa, the firms should take ownership via internal policies and procedures while their respective regulatory authorities and/or self-regulatory organizations, where appropriate and applicable, would impose boundaries and uphold governance through relevant laws, guidelines, and best practices.
Ming, thank you and I am sure things will evolve as we see progress in these technologies and newer use cases. Saving Changes...
Sergio Luis ConteHelping to create solutions for everyone| Worldwide based OrganizationsBuenos Aires, Argentina
Decisions are always make by human beings. AI just give the human being three scenarios (at least) with probabilities associated to them then the human being take the decision. Nothing different people is doing today. More dangerous is do not stay clear about what AI really is or to stay in "The Matrix" and trust in fake news or message like happens pay attention to whatsapp, X, etc, etc Saving Changes...
"Humanity has advanced, when it has advanced, not because it has been sober, responsible and cautious, but because it has been playful, rebellious and immature."
