Groups | Search | Server Info | Keyboard shortcuts | Login | Register

Groups > comp.misc > #28579

Re: Is this where we are going with AI?

Show all headers | View raw

Jim Jackson <jj@franjam.org.uk> wrote:
> On 2026-05-12, Computer Nerd Kev <not@telling.you.invalid> wrote:
>> Jim Jackson <jj@franjam.org.uk> wrote:
> ....snip....
>>> I reckon to use one of these tools well, you already need to be somewhat 
>>> of an expert in the area of discussion - so you can spot the 
>>> hallucinations and where it is off the mark. 
>>
>> Even then they still describe their hallucinations in dangerously
>> convincing ways, wrapping them with lots of irrelevent but
>> correct facts from elsewhere. More misleading than any response
>> from a human, who would never work that hard to sell such lies in
>> an unimportant online discussion.
> 
> An aspect That hadn't occurred to me! Thanks.
> 
> Though of course that ignores the trolls that often go out of their way 
> to mislead.

To me trolls are the people I meant who "sell such lies in an
unimportant online discussion". They might blabbler on a lot, but
in an unconvincing lazy way referencing few concrete facts. Unlike
the AI responses which tend to include lots of specific peripheral
facts even if their answer to the actual question is a complete
hallucination. If you count as trolls people who aren't lying, who
really believe and care about all the arguments they're making, but
just never stop arguing them, that might be different.

> Which begs the question: Are AIs trolls, but aren't aware of it?

The AI chatbots do seem to try and tell you what you want to hear
rather than pick arguments just to keep the discussion going like a
troll. But if they do get paid for advertising, the more engagement
they get, the more opportunity to reference sponsored
products/topics, so maybe their behaviour will gradually be tweaked
share that same troll-like motivation to just keep stringing users
along whatever it takes?

Mind you I wouldn't consider LLM AI chatbots to be "aware" of
this either way. Being aware of things is a whole other level of
technology, for better or worse.

-- 
__          __
#_ < |\| |< _#

Back to comp.misc | Previous | Next — Previous in thread | Next in thread | Find similar

Thread

Is this where we are going with AI? Jim Jackson <jj@franjam.org.uk> - 2026-05-12 18:26 +0000
  Re: Is this where we are going with AI? none <none@none.rip> - 2026-05-12 20:49 +0200
    Re: Is this where we are going with AI? oldernow <oldernow@dev.null> - 2026-05-13 00:07 +0000
      Re: Is this where we are going with AI? none <none@none.rip> - 2026-05-13 02:28 +0200
        Re: Is this where we are going with AI? oldernow <oldernow@dev.null> - 2026-05-13 02:58 +0000
  Re: Is this where we are going with AI? ram@zedat.fu-berlin.de (Stefan Ram) - 2026-05-12 18:56 +0000
    Re: Is this where we are going with AI? Jim Jackson <jj@franjam.org.uk> - 2026-05-12 19:19 +0000
  Re: Is this where we are going with AI? Richmond <dnomhcir@gmx.com> - 2026-05-12 20:10 +0100
    Re: Is this where we are going with AI? Jim Jackson <jj@franjam.org.uk> - 2026-05-12 19:17 +0000
      Re: Is this where we are going with AI? Richmond <dnomhcir@gmx.com> - 2026-05-12 20:32 +0100
  Re: Is this where we are going with AI? none <none@none.rip> - 2026-05-12 21:52 +0200
  Re: Is this where we are going with AI? not@telling.you.invalid (Computer Nerd Kev) - 2026-05-13 08:12 +1000
    Re: Is this where we are going with AI? Jim Jackson <jj@franjam.org.uk> - 2026-05-13 09:48 +0000
      Re: Is this where we are going with AI? oldernow <oldernow@dev.null> - 2026-05-13 12:18 +0000
      Re: Is this where we are going with AI? not@telling.you.invalid (Computer Nerd Kev) - 2026-05-14 08:52 +1000
  Re: Is this where we are going with AI? Adrian Caspersz <email@here.invalid> - 2026-05-13 20:48 +0100
    Re: Is this where we are going with AI? none <none@none.rip> - 2026-05-13 21:59 +0200
  Re: Is this where we are going with AI? Oregonian Haruspex <no_email@invalid.invalid> - 2026-05-13 21:46 +0000
    Re: Is this where we are going with AI? Jim Jackson <jj@franjam.org.uk> - 2026-05-14 08:52 +0000

csiph-web