Dr. Lucky Tran May 30, 2025
Including fake studies and using AI to generate a report should instantly disqualify anyone to lead a scientific agency.
Show threadRowan Brad Quni-GudzinasMay 30, 2025

@luckytran Do you have anyone in mind? I use AI to generate reports often. I just did...

https://doi.org/10.5281/zenodo.15555539

The #fake studies thing is problematic but that's a #human #editor not #fact checking.

Autaxys Research & Development Master Plan

Autaxys Research & Development Master Plan (v1.0 Draft - Approved for Save)

Zenodo
Show threadCassandrichMay 30, 2025
@QNFO @luckytran .. yay for telling on yourself, I guess? 🤷
Show threadRowan Brad Quni-Gudzinas
@dalias @luckytran since I do not know what prompted the OP, I wish not to take sides in someone else's issue. Reckless or fraudulent behavior is clearly unacceptable. But what I suspect is the case is one of #AI #hallucination and #human carelessness: like the "AI lawyer" who didn't bother to check that the cases #ChatGPT spit out were real.
May 30, 2025 at 8:53pmWeb
Show threadRowan Brad Quni-GudzinasMay 30, 2025
@dalias @luckytran ...if so, then my soapbox is one of growing pains: #AI #language models are becoming indispensable, like it or not. If the report at issue endangered lives or was a deliberate lie, so be it. Otherwise, we would do well to better #understand how an #LLM works to avoid such mistakes as writers and judge those we find more as typos than #plagiarism.
Show threadRowan Brad Quni-GudzinasMay 30, 2025

@dalias @luckytran What do I mean by "typo instead of plagiarism?" This was in a recent article I authored:

“Note on References: This work is a collaborative effort between the human author and an AI large language model (Google's Gemini). As a knowledge synthesis, the content is generated from a vast corpus of information, and the process of identifying and citing specific references is akin to reverse-engineering. While every attempt is made to ensure accuracy and avoid hallucination, ...

Show threadRowan Brad Quni-GudzinasMay 30, 2025

@dalias @luckytran

... the reference generation process fundamentally differs from conventional research, which typically starts with individual works. This approach may result in imperfections in the completeness or verifiability of all external links, as the primary goal is to synthesize coherent knowledge rather than to meticulously track every source in a traditional bibliographic manner.”

(Can't wait for the replies. The OP issue may be different but it strikes a chord heard regularly)

Show threadCassandrichMay 30, 2025
@QNFO @luckytran No they are not "indispensable". It's clear that you're here to be a dutiful cult member pushing that narrative, and the more you push it, the more slimy it all looks to the rest of us.
Show threadLisPiMay 31, 2025
@QNFO @luckytran @dalias > ...if so, then my soapbox is one of growing pains: #AI #language models are becoming indispensable, like it or not.

How have they done so? You seem to still be capable of writing.
Show threadPatrick O'BeirneMay 31, 2025
@lispi314
We're still capable of laziness.
Show threadLisPiMay 31, 2025
@PatrickOBeirne Not bothering to do one's work properly without fraudulent behavior & scholarly misconduct is a rather strong argument for not doing it at all. It'll do less harm.
Show threadFrank Bennett 🇯🇵May 30, 2025
@QNFO @dalias @luckytran Reaching for the Cliff Notes version of reality instead of the word processor is the original disqualifying mistake.
Show threadCassandrichMay 30, 2025
@QNFO @luckytran Using a slop generator and pretending you wrote the output without even being aware of the content is already scholarly misconduct.