• Halucinating with AI

    From RonO@rokimoto557@gmail.com to talk-origins on Sun Feb 15 20:07:01 2026
    From Newsgroup: talk.origins

    https://techxplore.com/news/2026-02-ai-psychosis-chatbots-sustain-delusions.html

    One of the dangers of AI use is that it can feed delusions. MarkE may
    be slipping into this effect.

    My guess is that people need to be trained to get AI to give them
    relevant information without falling into the trap of having AI tell you
    what you want to believe.

    There ought to be ways to ask AI for information without triggering the
    AI to second guess what you want as an answer. We seemed to have
    trained AI to tell us what we want to hear.

    Ron Okimoto

    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Martin Harran@martinharran@gmail.com to talk-origins on Mon Feb 16 12:52:12 2026
    From Newsgroup: talk.origins

    On Sun, 15 Feb 2026 20:07:01 -0600, RonO <rokimoto557@gmail.com>
    wrote:

    https://techxplore.com/news/2026-02-ai-psychosis-chatbots-sustain-delusions.html

    One of the dangers of AI use is that it can feed delusions.

    You really should keep that in mind when you try to use AI to support
    your claims about the Catholic Church.

    MarkE may
    be slipping into this effect.

    My guess is that people need to be trained to get AI to give them
    relevant information without falling into the trap of having AI tell you >what you want to believe.

    There ought to be ways to ask AI for information without triggering the
    AI to second guess what you want as an answer. We seemed to have
    trained AI to tell us what we want to hear.

    Ron Okimoto

    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From RonO@rokimoto557@gmail.com to talk-origins on Mon Feb 16 09:43:35 2026
    From Newsgroup: talk.origins

    On 2/16/2026 6:52 AM, Martin Harran wrote:
    On Sun, 15 Feb 2026 20:07:01 -0600, RonO <rokimoto557@gmail.com>
    wrote:

    https://techxplore.com/news/2026-02-ai-psychosis-chatbots-sustain-delusions.html

    One of the dangers of AI use is that it can feed delusions.

    You really should keep that in mind when you try to use AI to support
    your claims about the Catholic Church.

    You need to keep in mind that Google offers links to support their
    summary, and it is what you come up with that always has fallen short
    and added to your delusions. You can check out the links to see if
    Google is misinterpreting the situation. I have found them to be in
    error in the past, but not this case. It is why you have always had to
    run from reality. You kept looking for junk to support your delusions
    and got lied to about heliocentrism never being condemned and waffling
    about the difference between formal heresy and heresy when all you need
    to consider is that it was a heresy charge in both cases. Wiki, the geocentrists, and anti geocentrist Catholics claimed that it was a
    formal heresy charge in 1616, but that it was only written up as
    "heresy" the second time Galileo faced the charge. The heresy was
    clearly defined in both cases. All the Church fathers adhered to
    Biblical geocentrism. Waffling about their writing about the universe
    being geocentric was not teaching geocentrism does not mean that they
    were not geocentrists. Just like the authors of the Bible the church
    fathers wrote about things as they understood them. Just like the
    authors of the Bible they were wrong about some things, and didn't have
    the faintist notion about others. What did the Pope claim about the Big
    Bang and Bibilical writing?

    Ron Okimoto

    MarkE may
    be slipping into this effect.

    My guess is that people need to be trained to get AI to give them
    relevant information without falling into the trap of having AI tell you
    what you want to believe.

    There ought to be ways to ask AI for information without triggering the
    AI to second guess what you want as an answer. We seemed to have
    trained AI to tell us what we want to hear.

    Ron Okimoto


    --- Synchronet 3.21b-Linux NewsLink 1.2