Credibility of Artificial Intelligence Enabled Content - A Question Mark

I found the following two JAMA guidances concerning AI use; they basically say submitters and peer reviewers should be upfront on their use of AI when creating content or reviewing work. I feel these guidances are practically worthless, as I doubt someone who would use AI as a shortcut for actual research would have any qualms lying about it. The guidance for reviewers includes the following: “Please remember that you are ultimately responsible for all of the content of this review.”. However, this has little weight unless there are defined legal penalties for publishing improperly reviewed research.
In the past we could reasonably rely on peer review to verify the accuracy of research, but if reviewers can also use AI that’s no longer the case. I think the best way to ensure journals like JAMA remain credible is for the medical community to demand they do so, stressing the importance of the journals in edifying medical personnel around the globe, and the negative impacts that would inevitably result from the dissemination of incorrect information, including loss of human lives.

Guidance for Authors, Peer Reviewers, and Editors on Use of AI, Language Models, and Chatbots https://jamanetwork.com/journals/jama/fullarticle/2807956

Reporting Use of AI in Research and Scholarly Publication—JAMA Network Guidance
https://jamanetwork.com/journals/jama/fullarticle/2816213

Thanks for sharing, Deepa.

I believe AI as a tool can help people deliver better results. For instance if someone has a well-thought-out argument and uses ChatGPT to research to flesh out the argument, that could result in an argument that delivers better value

Sadly, there seem to be more people (especially on YouTube) boasting about how they can completely "outsource" their thinking to ChatGPT and other LLMs.

In my experience, blatant ChatGPT-generated responses are usually obvious with their hard-to-read blocks of text.

So I don't even bother wasting more time reading them

Thanks again

I think it's an absolute tragedy for students to use AI to generate their assignments. Thinking your way through the problem is how you generate those neural pathways where we develop our own mental models that provide deeper understandings of how things really work.

My professional writing process usually starts with many pages of notes that I then spend hours organizing, condensing, and explaining out-loud so I can build more mental pathways and hear how it "flows". I would love to use AI to organize my notes and fix long awkward sentences, but the underlying thoughts, and the progression of how I bring you from point A to B are still my own.

In the business world, when trying to solve problems one of our biggest obstacles is often that we have very little data to guide us. If you try to draw inferences from a very small dataset, you wind up with outputs that are very biased based on the small sample. I see that with PMI's Project Infinity where answers to my questions cite only 1 or 2 articles, and if it's one then the answer goes off track trying to build a longer answer with little relevant information.. To dig deeper, it becomes a matter of deciding what variables need to be investigated in greater depth to provide the more complete picture of cause and effect. To do that requires being grounded in the underlying principles.

AI can help people quickly gather and organize information on the current state of the art. If you want to expand the current body of knowledge however, that requires original insights. You can't get there purely focused on the rear-view mirror.

I so agree with you, and most do, I guess.

But then, I wonder if our content decaying over a period of time, and who is responsible? what legacy are we giving our generations from a credible information? Would love to know your thoughts.

Thanks again, Deepa. Enjoying this conversation

You asked: "I wonder if our content decaying over a period of time, and who is responsible? what legacy are we giving our generations from a credible information?"

Content will keep getting better as long as more people use AI intelligently. Poor quality content will always be generated by people too lazy to do their own thinking and fact-checking.

About the legacy to future generations, let's hope they learn that there's no way to avoid the rigours of actually doing our homework. Yes, we may be able to do things faster, but we still need to be clear about the outcomes we want and WHY. AI CANNOT do our own thinking

Thanks again, Deepa