Re: Re: ChatGPT Writing
By: jimmylogan to Rob Mccart on Sat Oct 04 2025 17:02:50

And I think if we stop calling it AI, which is technically
a misnomer, it might make it less frightening... It is
LLM - Language Learning Module - and has absolutely no
sentience behind it. It's not 'intellegent,' it is just
programmed to respond and such in a way that is comfortable
to us.

Thank you! I'm constantly trying to explain this to others. Unfortunetly, trying to explain LLM makes peoples eyes glaze over, so I can understand why marketing and media folks labeled it as AI.

---
þ Synchronet þ End Of The Line BBS - endofthelinebbs.com

Re: AI a Lie
By: Mortar to jimmylogan on Mon Oct 06 2025 11:25 am

Thank you! I'm constantly trying to explain this to others.
Unfortunetly, trying to explain LLM makes peoples eyes glaze over, so I can understand why marketing and media folks labeled it as AI.

AI isn't a terrible name. The "intelligence" is artificial. (not real intelligence)

That said, all modern "AI" is just a very advanced version of the system that predicts what word you're going to type next on your phone.

...I don't deserve this, but I have arthritis and I don't deserve that either

---
þ Synchronet þ War Ensemble BBS - The sport is war, total war - warensemble.com

And I think if we stop calling it AI, which is technically
a misnomer, it might make it less frightening... It is
LLM - Language Learning Module - and has absolutely no
sentience behind it. It's not 'intellegent,' it is just
programmed to respond and such in a way that is comfortable
to us.

It might be a misnomer, but I am not sure about "no" sentience. It has
been proven that AI/LLM is more likely than a human to "cheat" in order to
get the outcome it wants. Whether that is a sign of "some" sentience, or
if it is merely a sign that machines don't have ethics, is a subject for debate.


* SLMR 2.1a * ...a host of holy horrors to direct our aimless dance...
---
þ Synchronet þ CAPCITY2 * capcity2.synchro.net * Telnet/SSH:2022/Rlogin/HTTP

Re: AI a Lie
By: Dumas Walker to JIMMYLOGAN on Tue Oct 07 2025 09:02 am

been proven that AI/LLM is more likely than a human to "cheat" in order to get the outcome it wants. Whether that is a sign of "some" sentience, or
if it is merely a sign that machines don't have ethics, is a subject for debate.

Not sure sentience really has anything to do with it. What the computer knows is that it has an objective. If "cheating" allows it to achieve its objective faster, what is really stopping it? Some algorithm that says it won't cheat some X percent of the time?

---
þ Synchronet þ TIRED of waiting 2 hours for a taco? GO TO TACOPRONTO.bbs.io

Mortar wrote to jimmylogan <=-

Thank you! I'm constantly trying to explain this to others.
Unfortunetly, trying to explain LLM makes peoples eyes glaze over, so I can understand why marketing and media folks labeled it as AI.

I thought Machine Learning summed it up nicely.



--- MultiMail/Win v0.52
þ Synchronet þ .: realitycheckbbs.org :: scientia potentia est :.

Hello Dumas!

** On Tuesday 07.10.25 - 09:02, Dumas Walker wrote to JIMMYLOGAN:

It might be a misnomer, but I am not sure about "no"
sentience. It has been proven that AI/LLM is more likely
than a human to "cheat" in order to get the outcome it
wants. Whether that is a sign of "some" sentience, or if
it is merely a sign that machines don't have ethics, is a
subject for debate.

Nah.. no sentience. It's just reacting to what it was
programmed to do/choose.

--- OpenXP 5.0.64
* Origin: Ogg's Dovenet Point (723:320/1.9)
* Synchronet * CAPCITY2 * capcity2.synchro.net * Telnet/SSH:2022/Rlogin/HTTP

By: jimmylogan to Rob Mccart on Sat Oct 04 2025 17:02:50

J > > And I think if we stop calling it AI, which is technically
> > a misnomer, it might make it less frightening... It is
> > LLM - Language Learning Module - and has absolutely no
> > sentience behind it. It's not 'intellegent,' it is just
> > programmed to respond and such in a way that is comfortable
> > to us.

Thank you! I'm constantly trying to explain this to others. Unfortunetly, t

>ng to explain LLM makes peoples eyes glaze over, so I can understand why mark
>ng and media folks labeled it as AI.

To add a bit to what I said last night, I don't want to think of these
systems as being truly sentient but a story that cropped up a while back
would make you stop and think.

As I said, they often tend to make up an answer if they can't find
one, trying too hard to 'please', but it came out a while back
that one 'AI' system that they had lots of things for it to work
on was given limited time for whatever it was working on at the
moment.. Things seemed a bit off a while later and they checked
and found that when the system couldn't complete a job it was
working on in the allotted time, it actualy rewrote part of its
own programming to give itself more time to complete the work.

I'm not sure what you'd call that but it sounds rather 'intelligent'
to me, almost worrisome since it is overriding the orders it had
been given by the human users..

I'm not sure where the name LLM came from since it's not just
about language, although the 'learning' part of it might be a
big part of that.. B)

---
þ SLMR Rob þ Caution... Tagline under construction...
þ Synchronet þ CAPCITY2 * capcity2.synchro.net * Telnet/SSH:2022/Rlogin/HTTP

Thank you! I'm constantly trying to explain this to others.
Unfortunetly, trying to explain LLM makes peoples eyes glaze over, so I
can understand why marketing and media folks labeled it as AI.

AI isn't a terrible name. The "intelligence" is artificial. (not real intel

>ence)

That said, all modern "AI" is just a very advanced version of the system tha

>redicts what word you're going to type next on your phone.

That last is a relly good example of what an advanced LLM could do, but
a lot of AI systems seem able to invent things on their own and make
decisions against their original programming/orders.

I think it's come a long way in recent years and must do some
pretty spectacular things or we wouldn't be spending many
$Billions to create and support more of them..

Obviously some will be much more complex than others and able
to do a lot more.

---
þ SLMR Rob þ Please hold... All our Taglines are busy at the moment
þ Synchronet þ CAPCITY2 * capcity2.synchro.net * Telnet/SSH:2022/Rlogin/HTTP

Hello Rob!

...was given limited time for whatever it was working on at the
moment.. Things seemed a bit off a while later and they checked
and found that when the system couldn't complete a job it was
working on in the allotted time, it actualy rewrote part of its
own programming to give itself more time to complete the work.

Nah.. it didn't rewrite anything. It's totally possible that
the original code failed to implement a hard fast rule when
time runs out thus allowing the process to continue.


--- OpenXP 5.0.64
* Origin: Ogg's Dovenet Point (723:320/1.9)
* Synchronet * CAPCITY2 * capcity2.synchro.net * Telnet/SSH:2022/Rlogin/HTTP

I'm not sure what you'd call that but it sounds rather 'intelligent'
to me, almost worrisome since it is overriding the orders it had
been given by the human users..

I'm not sure where the name LLM came from since it's not just
about language, although the 'learning' part of it might be a
big part of that.. B)

IMHO, if it is "learning" (which they are supposed to do), it might not be sentient, but it has the capacity to become intelligent enough to take
actions that have previously been available to only those of use who are sentient.


* SLMR 2.1a * Southern Serves the South
---
þ Synchronet þ CAPCITY2 * capcity2.synchro.net * Telnet/SSH:2022/Rlogin/HTTP

...was given limited time for whatever it was working on at the
moment.. Things seemed a bit off a while later and they checked
and found that when the system couldn't complete a job it was
working on in the allotted time, it actualy rewrote part of its
own programming to give itself more time to complete the work.

Nah.. it didn't rewrite anything. It's totally possible that

>the original code failed to implement a hard fast rule when
>time runs out thus allowing the process to continue.

I decided to have another look for that original story because
I figured if it was a common problem then it wouldn't have been
newsworthy enough to get on a National News broadcast..

So, this may be a case of both of us being right at some level
but the original story involved Sakana AI and it mentioned the
potential risks related to AI autonomy when their AI 'Attempted'
to modify its own code to extend the runtime of it's experiments,
which they said could lead to unexpected behaviors and challenges
in control.. and they go on to needing more robust safety
protocols and isolating AI systems from critical infrastructure
to prevent unintended consequences.

So the battle begins.. Every time people try to rein in what
their AI can do on its own, it will try to find a way around
that.. The old better mouse trap = better mouse problem.. B)

---
þ SLMR Rob þ click...click...click..Damn... out of taglines!
þ Synchronet þ CAPCITY2 * capcity2.synchro.net * Telnet/SSH:2022/Rlogin/HTTP

I'm not sure where the name LLM came from since it's not just

>> about language, although the 'learning' part of it might be a
>> big part of that.. B)

IMHO, if it is "learning" (which they are supposed to do), it might not be

>sentient, but it has the capacity to become intelligent enough to take
>actions that have previously been available to only those of use who are
>sentient.

My niece is a teacher and she tried something new this year for preparing report cards for her students. She estimates that it takes her about
60 hours to come up with all the comments required to add to the report
cards. It is expected there will be detailed comments for every student.
No 'boilerplate' allowed..

She got an AI involved where she could put in a few comments and the AI
took her comments and the marks and such and apparently amazed her and
everyone she showed it to how it came up with a report that was at least
as good as she'd have come up with in a matter of seconds.

But in this case it is using a little new information and making
a modified copy of typical old comments, so it's not so much
intelligence as manipulating data into an expected format.

---
þ SLMR Rob þ (A)bort, (R)etry, (U)se a better tagline
þ Synchronet þ CAPCITY2 * capcity2.synchro.net * Telnet/SSH:2022/Rlogin/HTTP

DaiTengu wrote to Mortar <=-

Re: AI a Lie
By: Mortar to jimmylogan on Mon Oct 06 2025 11:25 am

Thank you! I'm constantly trying to explain this to others.
Unfortunetly, trying to explain LLM makes peoples eyes glaze over, so I can understand why marketing and media folks labeled it as AI.

AI isn't a terrible name. The "intelligence" is artificial. (not real intelligence)

That said, all modern "AI" is just a very advanced version of the
system that predicts what word you're going to type next on your phone.

Agreed. I think AI has taken off because it was called that in
sci-fi.

Open the pod bay doors HAL - sorry Dave, but I can't do that.

HAL is not 'intelligent' but sounds like 'he' is making decisions.



... Xerox Alto was the thing. Anything after we use is just a mere copy.
--- MultiMail/Mac v0.52
þ Synchronet þ Digital Distortion: digitaldistortionbbs.com

Dumas Walker wrote to JIMMYLOGAN <=-

And I think if we stop calling it AI, which is technically
a misnomer, it might make it less frightening... It is
LLM - Language Learning Module - and has absolutely no
sentience behind it. It's not 'intellegent,' it is just
programmed to respond and such in a way that is comfortable
to us.

It might be a misnomer, but I am not sure about "no" sentience. It has been proven that AI/LLM is more likely than a human to "cheat" in order
to get the outcome it wants. Whether that is a sign of "some"
sentience, or if it is merely a sign that machines don't have ethics,
is a subject for debate.

I've not lookd into that 'proof' - but by sentience I mean self
aware, not just programmed to know, but KNOW. :-)


... Ensign Pillsbury? He's BREAD, Jim
--- MultiMail/Mac v0.52
þ Synchronet þ Digital Distortion: digitaldistortionbbs.com

phigan wrote to Dumas Walker <=-

Re: AI a Lie
By: Dumas Walker to JIMMYLOGAN on Tue Oct 07 2025 09:02 am

been proven that AI/LLM is more likely than a human to "cheat" in order to get the outcome it wants. Whether that is a sign of "some" sentience, or
if it is merely a sign that machines don't have ethics, is a subject for debate.

Not sure sentience really has anything to do with it. What the computer knows is that it has an objective. If "cheating" allows it to achieve
its objective faster, what is really stopping it? Some algorithm that
says it won't cheat some X percent of the time?

Hmm... Good point. So a computer is programmed to do something, like
solve a maze. Unless you tell it NOT to go through a particular
wall, maybe it will.

Okay, not a great analogy, but maybe that gets my point across. :-)



... Tolkien is hobbit-forming.
--- MultiMail/Mac v0.52
þ Synchronet þ Digital Distortion: digitaldistortionbbs.com