From Newsgroup: talk.origins

https://www.cbsnews.com/news/ai-hallucinate-citations-medial-research/

A study published in Lancet found 4,000 medical publications with made
up citations. Apparently the medical researchers are using AI to
generate their papers and the AI will fabricate citations needed. I am
glad that I no longer peer review papers. Once I retired I resigned
from my review position for Poultry Science (I was on some type of 5
year editorial review thing that they had asked me to sign on for
off-and-on for the last 30 years) and I did not give the other journals
my home email. Over 20 years ago I met a professional paper writer at
the Plant Animal genome meeting in San Diego. He claimed that he had a position at a medical college, and that it was his job to write up the material that the doctors would produce and get it published. He
claimed that a lot of the medical doctors did not know how to write a
research publication, and sometimes did not know if what they had done
was even publishable. It was his job to organize what they produced
into something that could be published. Apparently AI is being used for
that job, and for some reason it can resort to making up fake citations.

I would not check every citation. I would only check them if they were important to the conclusions of the paper, and I had no knowledge of the publication cited. Only a couple of times did I find that the authors
had cited the wrong paper (the paper cited existed, but did not support
what was claimed) in all cases where they had cited the wrong paper,
they actually had the correct citation as a reference, but did not cite
it where it should have been cited. So it seemed to be some type of
honest mistake. The AI is making up references that do not exist. Does
that mean that it can also make inferences that cannot be made? We were starting to use AI to sift through complex data sets and tell us things
that might possibly be gleaned from the data. It would be tragic if the
AI could start making up connections that do not exist, just to tell us
what we want to hear. It would be as bad as having the ID perps
interpret the data.

Ron Okimoto

--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: talk.origins

On 5/14/26 08:01, RonO wrote:

https://www.cbsnews.com/news/ai-hallucinate-citations-medial-research/

A study published in Lancet found 4,000 medical publications with made
up citations.-a Apparently the medical researchers are using AI to
generate their papers and the AI will fabricate citations needed.-a I am glad that I no longer peer review papers.-a Once I retired I resigned
from my review position for Poultry Science (I was on some type of 5
year editorial review thing that they had asked me to sign on for
off-and-on for the last 30 years) and I did not give the other journals
my home email.-a Over 20 years ago I met a professional paper writer at
the Plant Animal genome meeting in San Diego.-a He claimed that he had a position at a medical college, and that it was his job to write up the material that the doctors would produce and get it published.-a He
claimed that a lot of the medical doctors did not know how to write a research publication, and sometimes did not know if what they had done
was even publishable.-a It was his job to organize what they produced
into something that could be published.-a Apparently AI is being used for that job, and for some reason it can resort to making up fake citations.

I would not check every citation.-a I would only check them if they were important to the conclusions of the paper, and I had no knowledge of the publication cited.-a Only a couple of times did I find that the authors
had cited the wrong paper (the paper cited existed, but did not support
what was claimed) in all cases where they had cited the wrong paper,
they actually had the correct citation as a reference, but did not cite
it where it should have been cited.-a So it seemed to be some type of
honest mistake.-a The AI is making up references that do not exist.-a Does that mean that it can also make inferences that cannot be made?-a We were starting to use AI to sift through complex data sets and tell us things
that might possibly be gleaned from the data.-a It would be tragic if the
AI could start making up connections that do not exist, just to tell us
what we want to hear.-a It would be as bad as having the ID perps
interpret the data.

Ron Okimoto

This is from CBS news (you know an information source for
Deltas and Epsilons (you know, people who have had their
brain growth stunted before decanting) thus it must be
REALLY true).



--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: talk.origins

On 2026-05-15 2:50 p.m., x3 wrote:

On 5/14/26 08:01, RonO wrote:

https://www.cbsnews.com/news/ai-hallucinate-citations-medial-research/

A study published in Lancet found 4,000 medical publications with made
up citations.-a Apparently the medical researchers are using AI to
generate their papers and the AI will fabricate citations needed.-a I
am glad that I no longer peer review papers.-a Once I retired I
resigned from my review position for Poultry Science (I was on some
type of 5 year editorial review thing that they had asked me to sign
on for off-and-on for the last 30 years) and I did not give the other
journals my home email.-a Over 20 years ago I met a professional paper
writer at the Plant Animal genome meeting in San Diego.-a He claimed
that he had a position at a medical college, and that it was his job
to write up the material that the doctors would produce and get it
published.-a He claimed that a lot of the medical doctors did not know
how to write a research publication, and sometimes did not know if
what they had done was even publishable.-a It was his job to organize
what they produced into something that could be published.-a Apparently
AI is being used for that job, and for some reason it can resort to
making up fake citations.

I would not check every citation.-a I would only check them if they
were important to the conclusions of the paper, and I had no knowledge
of the publication cited.-a Only a couple of times did I find that the
authors had cited the wrong paper (the paper cited existed, but did
not support what was claimed) in all cases where they had cited the
wrong paper, they actually had the correct citation as a reference,
but did not cite it where it should have been cited.-a So it seemed to
be some type of honest mistake.-a The AI is making up references that
do not exist.-a Does that mean that it can also make inferences that
cannot be made?-a We were starting to use AI to sift through complex
data sets and tell us things that might possibly be gleaned from the
data.-a It would be tragic if the AI could start making up connections
that do not exist, just to tell us what we want to hear.-a It would be
as bad as having the ID perps interpret the data.

Ron Okimoto

This is from CBS news (you know an information source for
Deltas and Epsilons (you know, people who have had their
brain growth stunted before decanting) thus it must be
REALLY true).

yes, what with Lancet being such a fringe source for medical information. /s
--
--
Don Cates ("he's a cunning rascal" PN)

--- Synchronet 3.22a-Linux NewsLink 1.2