Us academics, and journal editors, have had frequent conversations lately about the role of AI/LLM in scholarly writing. This article in The Conversation provides some useful perspectives - and it covers issues such as credit for the writing/ideas, bias, hallucination - and of course the challenge of enforcement if any bans are made. And it ends with a good observation that the real goal should be to prevent sloppiness and maintain good quality. Would love to hear all your thoughts on this. https://lnkd.in/gxtfKPub #academia #journals #articles #genAI
Interestingly, LinkedIn just tried to offer me a premium service to have AI write my response. No thank you, but I am not sure I want an ethical prohibition. "Don't be sloppy,” is a good principle. On the flip side, common misspellings could be a subtle clue AI was not involved, so a norm of minor sloppyness could become a helpful principal😉 to find where the humans are.
Shiva Shekhar
6mo
It’s hard to do that if the papers are 50 pages long (generated by AI) and reviewers are not given much time/credit for their work. Then it just becomes a spam problem and everyone is worse off.
Venture Fellow @ Laconia | Consultant @ Innovation Strategy | Investment Analyst | MBA, Finance & Strategy
7moWhen I was worked in a lab characterizing novel high coordinate metal-metal complexes, the paper was a way to celebrate the efforts that went in to the research. It was the feather in the cap where researchers could put their signature on their findings. There’s also a great deal of garbage that gets pumped out and circulated, like that paper a year or two ago that the media picked up on (despite this claim being made several times a year for the last decade). I agree with the article’s conclusion, which is an issue that’s plagued academic journals for as long as I’ve been reading them. But that is a problem that’s compounded by an onslaught of submissions enabled by LLMs. I don’t think LLMs are particularly useful to people who are already talented communicators. Rather, they enable poor communicators to become better / faster. I can imagine benefits and drawbacks to that. I look forward to discriminative AI models that will detect generative AI. At some point maybe we’ll just have a layer in society of AI models engaged in virtual combat to counteract each other. Seems like as efficient a use of energy as bitcoin.