No Linux? No problem. Just get AI to hallucinate it for you
https://arstechnica.com/information-technology/2022/12/openais-new-chatbot-can-hallucinate-a-linux-shell-or-calling-a-bbs/

Worries me more than a bit..

On 12/6/2022 12:05 AM, Jan Panteltje wrote:
> No Linux? No problem. Just get AI to hallucinate it for you
> https://arstechnica.com/information-technology/2022/12/openais-new-chatbot-can-hallucinate-a-linux-shell-or-calling-a-bbs/
>
> Worries me more than a bit..

"Degrave found that the simulation goes surprisingly deep."

The mistake is believing it's actually simulating anything, it isn't.
There isn't any logically consistent virtual machine operating under the
hood, so commands of sufficient complexity will surely produce nonsense
eventually.

On 12/6/2022 12:05 AM, Jan Panteltje wrote:
> No Linux? No problem. Just get AI to hallucinate it for you
> https://arstechnica.com/information-technology/2022/12/openais-new-chatbot-can-hallucinate-a-linux-shell-or-calling-a-bbs/
>
> Worries me more than a bit..

thankfully it has close to no ability to design circuits - try entering
"Please show an ASCII schematic of an op amp circuit with a gain of 2 in
the non-inverting configuration" - when I tried that it went into an
infinite loop drawing pipes like mad, but it comes up with some
interesting ideas:

<https://imgur.com/a/L3INEHH>

Like I guess you could crudely represent how big a resistor was that way..

On a sunny day (Tue, 6 Dec 2022 08:33:26 -0500) it happened bitrex
<user@example.net> wrote in <GwHjL.2561$vBI8.1335@fx15.iad>:

>On 12/6/2022 12:05 AM, Jan Panteltje wrote:
>> No Linux? No problem. Just get AI to hallucinate it for you
>> https://arstechnica.com/information-technology/2022/12/openais-new-chatbot-can-hallucinate-a-linux-shell-or-calling-a-bbs/
>>
>> Worries me more than a bit..
>
>
>"Degrave found that the simulation goes surprisingly deep."
>
>The mistake is believing it's actually simulating anything, it isn't.
>There isn't any logically consistent virtual machine operating under the
>hood, so commands of sufficient complexity will surely produce nonsense
>eventually.

But is not that the same with us? ;-)

On 12/6/2022 10:59 AM, Jan Panteltje wrote:
> On a sunny day (Tue, 6 Dec 2022 08:33:26 -0500) it happened bitrex
> <user@example.net> wrote in <GwHjL.2561$vBI8.1335@fx15.iad>:
>
>> On 12/6/2022 12:05 AM, Jan Panteltje wrote:
>>> No Linux? No problem. Just get AI to hallucinate it for you
>>> https://arstechnica.com/information-technology/2022/12/openais-new-chatbot-can-hallucinate-a-linux-shell-or-calling-a-bbs/
>>>
>>> Worries me more than a bit..
>>
>>
>> "Degrave found that the simulation goes surprisingly deep."
>>
>> The mistake is believing it's actually simulating anything, it isn't.
>> There isn't any logically consistent virtual machine operating under the
>> hood, so commands of sufficient complexity will surely produce nonsense
>> eventually.
>
> But is not that the same with us? ;-)

One way for it to fail the Turing test is to ask it a sequence of
questions of escalating complexity and note that it never says "Jeez, I
don't know" but will always make an attempt to sound "smart" and like it
knows the answer, even if it's finally just rephrasing the question.

So yeah I guess it is like Usenet discussions in that respect..

On a sunny day (Tue, 6 Dec 2022 09:01:27 -0500) it happened bitrex
<user@example.net> wrote in <XWHjL.61$mmyc.22@fx37.iad>:

>On 12/6/2022 12:05 AM, Jan Panteltje wrote:
>> No Linux? No problem. Just get AI to hallucinate it for you
>> https://arstechnica.com/information-technology/2022/12/openais-new-chatbot-can-hallucinate-a-linux-shell-or-calling-a-bbs/
>>
>> Worries me more than a bit..
>
>thankfully it has close to no ability to design circuits - try entering
>"Please show an ASCII schematic of an op amp circuit with a gain of 2 in
>the non-inverting configuration" - when I tried that it went into an
>infinite loop drawing pipes like mad, but it comes up with some
>interesting ideas:
>
><https://imgur.com/a/L3INEHH>
>
>Like I guess you could crudely represent how big a resistor was that way..

Interesting,
Sure if it had drawn molecules or atoms and not just bars
it makes a lot of sense,

But even 'pipes', at some level an analogy between the flow of electricity
and the flow of water goes a long way.

I remember my electricity experiments and gaining understanding as a very small kid
just that you needed 2 wires to connect a light bulb to a battery...
like water flowing - circulating.

AI is still learning :-)

On 12/6/2022 11:08 AM, Jan Panteltje wrote:
> On a sunny day (Tue, 6 Dec 2022 09:01:27 -0500) it happened bitrex
> <user@example.net> wrote in <XWHjL.61$mmyc.22@fx37.iad>:
>
>> On 12/6/2022 12:05 AM, Jan Panteltje wrote:
>>> No Linux? No problem. Just get AI to hallucinate it for you
>>> https://arstechnica.com/information-technology/2022/12/openais-new-chatbot-can-hallucinate-a-linux-shell-or-calling-a-bbs/
>>>
>>> Worries me more than a bit..
>>
>> thankfully it has close to no ability to design circuits - try entering
>> "Please show an ASCII schematic of an op amp circuit with a gain of 2 in
>> the non-inverting configuration" - when I tried that it went into an
>> infinite loop drawing pipes like mad, but it comes up with some
>> interesting ideas:
>>
>> <https://imgur.com/a/L3INEHH>
>>
>> Like I guess you could crudely represent how big a resistor was that way..
>
> Interesting,
> Sure if it had drawn molecules or atoms and not just bars
> it makes a lot of sense,
>
> But even 'pipes', at some level an analogy between the flow of electricity
> and the flow of water goes a long way.
>
> I remember my electricity experiments and gaining understanding as a very small kid
> just that you needed 2 wires to connect a light bulb to a battery...
> like water flowing - circulating.
>
> AI is still learning :-)
>

We're to the Searle's Chinese Room arguments as to whether it actually
learns or "understands" anything:

<https://en.wikipedia.org/wiki/Chinese_room>

The biological naturalists would say that it understands nothing about
circuits or electricity and has no ability to, while the functionalists
would say if it quacks like a duck, it's a duck.

But it's still pretty far from quacking like an EE, much less a duck, so
I think the question as to whether our own "consciousnesses" is just an
emergent property of Markov-chaining lexicons together will be an open
one for some time

On 12/6/2022 11:18 AM, bitrex wrote:
> On 12/6/2022 11:08 AM, Jan Panteltje wrote:
>> On a sunny day (Tue, 6 Dec 2022 09:01:27 -0500) it happened bitrex
>> <user@example.net> wrote in <XWHjL.61$mmyc.22@fx37.iad>:
>>
>>> On 12/6/2022 12:05 AM, Jan Panteltje wrote:
>>>> No Linux? No problem. Just get AI to hallucinate it for you
>>>>
>>>> https://arstechnica.com/information-technology/2022/12/openais-new-chatbot-can-hallucinate-a-linux-shell-or-calling-a-bbs/
>>>>
>>>> Worries me more than a bit..
>>>
>>> thankfully it has close to no ability to design circuits - try entering
>>> "Please show an ASCII schematic of an op amp circuit with a gain of 2 in
>>> the non-inverting configuration" - when I tried that it went into an
>>> infinite loop drawing pipes like mad, but it comes up with some
>>> interesting ideas:
>>>
>>> <https://imgur.com/a/L3INEHH>
>>>
>>> Like I guess you could crudely represent how big a resistor was that
>>> way..
>>
>> Interesting,
>> Sure if it had drawn molecules or atoms and not just bars
>> it makes a lot of sense,
>>
>> But even 'pipes', at some level an analogy between the flow of
>> electricity
>> and the flow of water goes a long way.
>>
>> I remember my electricity experiments and gaining understanding as a
>> very small kid
>> just that you needed 2 wires to connect a light bulb to a battery...
>> like water flowing - circulating.
>>
>> AI is still learning :-)
>>
>
> We're to the Searle's Chinese Room arguments as to whether it actually
> learns or "understands" anything:
>
> <https://en.wikipedia.org/wiki/Chinese_room>
>
> The biological naturalists would say that it understands nothing about
> circuits or electricity and has no ability to, while the functionalists
> would say if it quacks like a duck, it's a duck.
>
> But it's still pretty far from quacking like an EE, much less a duck, so
> I think the question as to whether our own "consciousnesses" is just an
> emergent property of Markov-chaining lexicons together will be an open
> one for some time
>

My intuition is that the development of consciousness requires agency; a
universe of some type to interact with in a meaningful way.
Consciousness without agency sounds like a type of hell to me, perhaps
there are physical firewalls against conscious minds constructing those
kinds of hells for other conscious minds.

I hope there are, anyway.

On Tuesday, 6 December 2022 at 17:30:02 UTC+1, bitrex wrote:
> On 12/6/2022 11:18 AM, bitrex wrote:
> > On 12/6/2022 11:08 AM, Jan Panteltje wrote:
> >> On a sunny day (Tue, 6 Dec 2022 09:01:27 -0500) it happened bitrex
> >> <us...@example.net> wrote in <XWHjL.61$mmy...@fx37.iad>:
> >>
> >>> On 12/6/2022 12:05 AM, Jan Panteltje wrote:
> >>>> No Linux? No problem. Just get AI to hallucinate it for you
> >>>>
> >>>> https://arstechnica.com/information-technology/2022/12/openais-new-chatbot-can-hallucinate-a-linux-shell-or-calling-a-bbs/
> >>>>
> >>>> Worries me more than a bit..
> >>>
> >>> thankfully it has close to no ability to design circuits - try entering
> >>> "Please show an ASCII schematic of an op amp circuit with a gain of 2 in
> >>> the non-inverting configuration" - when I tried that it went into an
> >>> infinite loop drawing pipes like mad, but it comes up with some
> >>> interesting ideas:
> >>>
> >>> <https://imgur.com/a/L3INEHH>
> >>>
> >>> Like I guess you could crudely represent how big a resistor was that
> >>> way..
> >>
> >> Interesting,
> >> Sure if it had drawn molecules or atoms and not just bars
> >> it makes a lot of sense,
> >>
> >> But even 'pipes', at some level an analogy between the flow of
> >> electricity
> >> and the flow of water goes a long way.
> >>
> >> I remember my electricity experiments and gaining understanding as a
> >> very small kid
> >> just that you needed 2 wires to connect a light bulb to a battery...
> >> like water flowing - circulating.
> >>
> >> AI is still learning :-)
> >>
> >
> > We're to the Searle's Chinese Room arguments as to whether it actually
> > learns or "understands" anything:
> >
> > <https://en.wikipedia.org/wiki/Chinese_room>
> >
> > The biological naturalists would say that it understands nothing about
> > circuits or electricity and has no ability to, while the functionalists
> > would say if it quacks like a duck, it's a duck.
> >
> > But it's still pretty far from quacking like an EE, much less a duck, so
> > I think the question as to whether our own "consciousnesses" is just an
> > emergent property of Markov-chaining lexicons together will be an open
> > one for some time
> >
> My intuition is that the development of consciousness requires agency; a
> universe of some type to interact with in a meaningful way.
> Consciousness without agency sounds like a type of hell to me, perhaps
> there are physical firewalls against conscious minds constructing those
> kinds of hells for other conscious minds.
>
> I hope there are, anyway.
visit secondlife.com one day

On 12/6/2022 12:05 AM, Jan Panteltje wrote:
> No Linux? No problem. Just get AI to hallucinate it for you
> https://arstechnica.com/information-technology/2022/12/openais-new-chatbot-can-hallucinate-a-linux-shell-or-calling-a-bbs/
>
> Worries me more than a bit..

It should. Imagine an AI assigned to each of us. Tracking us. Monitoring our behaviour. Knowing what we buy. What advertising we see. What we post in forums, product reviews, and emails. And then using that information to manipulate us. An AI that chooses what advertising we see. An AI that chooses what news headlines we see. What entertainment were directed to. How much of advertising is about the product and how much just manipulation? Pulling the strings of self image, desires, weaknesses for gambling, etc. How much is news and entertainment pushing an agenda? What if an AI could direct us to the ones that are written the right way to manipulate us. Pulling our strings. Turning us into puppets. Stealing our souls. And Google, Apple, Microsoft, etc. can say, "It's all encrypted. No person ever sees it. It's completely private."

Of course, I'm too smart for it to work on me.

On 2022-12-06 12:26, Wanderer@noemail.com wrote:
> On 12/6/2022 12:05 AM, Jan Panteltje wrote:
>> No Linux? No problem. Just get AI to hallucinate it for you
>> https://arstechnica.com/information-technology/2022/12/openais-new-chatbot-can-hallucinate-a-linux-shell-or-calling-a-bbs/
>>
>>
>>
Worries me more than a bit..
>
> It should. Imagine an AI assigned to each of us. Tracking us.
> Monitoring our behaviour. Knowing what we buy. What advertising we
> see. What we post in forums, product reviews, and emails. And then
> using that information to manipulate us. An AI that chooses what
> advertising we see. An AI that chooses what news headlines we see.
> What entertainment were directed to. How much of advertising is about
> the product and how much just manipulation? Pulling the strings of
> self image, desires, weaknesses for gambling, etc. How much is news
> and entertainment pushing an agenda? What if an AI could direct us to
> the ones that are written the right way to manipulate us. Pulling our
> strings. Turning us into puppets. Stealing our souls. And Google,
> Apple, Microsoft, etc. can say, "It's all encrypted. No person ever
> sees it. It's completely private."
>
> Of course, I'm too smart for it to work on me.
>

Oh? Is that why people keep reacting to entities like 'a a' and
'Fred Blogs'? I wonder how many bots keep this forum busy with
inanities. I kill-filter over 90% these days.

Jeroen Belleman

On 12/6/2022 12:26 PM, Wanderer@noemail.com wrote:
> On 12/6/2022 12:05 AM, Jan Panteltje wrote:
>> No Linux? No problem. Just get AI to hallucinate it for you
>> https://arstechnica.com/information-technology/2022/12/openais-new-chatbot-can-hallucinate-a-linux-shell-or-calling-a-bbs/
>>
>> Worries me more than a bit..
>
> It should. Imagine an AI assigned to each of us. Tracking us. Monitoring our behaviour. Knowing what we buy. What advertising we see. What we post in forums, product reviews, and emails. And then using that information to manipulate us. An AI that chooses what advertising we see. An AI that chooses what news headlines we see. What entertainment were directed to. How much of advertising is about the product and how much just manipulation? Pulling the strings of self image, desires, weaknesses for gambling, etc. How much is news and entertainment pushing an agenda? What if an AI could direct us to the ones that are written the right way to manipulate us. Pulling our strings. Turning us into puppets. Stealing our souls. And Google, Apple, Microsoft, etc. can say, "It's all encrypted. No person ever sees it. It's completely private."
>
> Of course, I'm too smart for it to work on me.

Imagine if you could read your secret file and it said "Not worth the
effort to manipulate."

I expect many Americans would be offended if that was the analysis.

On a sunny day (Tue, 6 Dec 2022 11:29:54 -0500) it happened bitrex
<user@example.net> wrote in <66KjL.4343$Tcw8.2329@fx10.iad>:

>On 12/6/2022 11:18 AM, bitrex wrote:
>> On 12/6/2022 11:08 AM, Jan Panteltje wrote:
>>> On a sunny day (Tue, 6 Dec 2022 09:01:27 -0500) it happened bitrex
>>> <user@example.net> wrote in <XWHjL.61$mmyc.22@fx37.iad>:
>>>
>>>> On 12/6/2022 12:05 AM, Jan Panteltje wrote:
>>>>> No Linux? No problem. Just get AI to hallucinate it for you
>>>>>
>>>>> https://arstechnica.com/information-technology/2022/12/openais-new-chatbot-can-hallucinate-a-linux-shell-or-calling-a-bbs/
>>>>>
>>>>> Worries me more than a bit..
>>>>
>>>> thankfully it has close to no ability to design circuits - try entering
>>>> "Please show an ASCII schematic of an op amp circuit with a gain of 2 in
>>>> the non-inverting configuration" - when I tried that it went into an
>>>> infinite loop drawing pipes like mad, but it comes up with some
>>>> interesting ideas:
>>>>
>>>> <https://imgur.com/a/L3INEHH>
>>>>
>>>> Like I guess you could crudely represent how big a resistor was that
>>>> way..
>>>
>>> Interesting,
>>> Sure if it had drawn molecules or atoms and not just bars
>>> it makes a lot of sense,
>>>
>>> But even 'pipes', at some level an analogy between the flow of
>>> electricity
>>> and the flow of water goes a long way.
>>>
>>> I remember my electricity experiments and gaining understanding as a
>>> very small kid
>>> just that you needed 2 wires to connect a light bulb to a battery...
>>> like water flowing - circulating.
>>>
>>> AI is still learning :-)
>>>
>>
>> We're to the Searle's Chinese Room arguments as to whether it actually
>> learns or "understands" anything:
>>
>> <https://en.wikipedia.org/wiki/Chinese_room>
>>
>> The biological naturalists would say that it understands nothing about
>> circuits or electricity and has no ability to, while the functionalists
>> would say if it quacks like a duck, it's a duck.
>>
>> But it's still pretty far from quacking like an EE, much less a duck, so
>> I think the question as to whether our own "consciousnesses" is just an
>> emergent property of Markov-chaining lexicons together will be an open
>> one for some time
>>
>
>
>My intuition is that the development of consciousness requires agency; a
>universe of some type to interact with in a meaningful way.
>Consciousness without agency sounds like a type of hell to me, perhaps
>there are physical firewalls against conscious minds constructing those
>kinds of hells for other conscious minds.
>
>I hope there are, anyway.

I have come to the belief that we (life) are just a simple self-assembly
of elementary particles, like an electron starts orbiting a nucleus
and molecules then form later and reactions start to happen between those
molecules.. and there we are.
All matter is 'conscious' in a way, moving an electron here a little bit later
affects an electron at the other end of the universe,
If a doctor test your eye-reflex to see if you are still conscious he is just testing
a system with a sensor that can move something, like a sunshade with a light sensor that can close
and open.
If that sunshade system has a speech output module it can say: "Sunny, I will close."
If the sensor signal or any other part of the system is going via some other module
that interrupts it, then we would say "its not conscious anymore'
Nothing mystical..
All is interacting.. ever more complex systems [life forms if you will] form.
What goes goes...

Long ago a guy at work who did a study of philosophy let me read one of his books.
Funny how you can go into reasoning in circles...
'I think so I exist' sort of stuff, not for me,
He left, went to an other country IIRC.

Question remains of course 'what is that electron (and all other particles it is made of)'
CERN will work on that ;-) But due to power cuts they cannot run full speed I've heard.

We are, after all, just like an ant creeping up a wall, not knowing
why the wall is there, what it is made of, who was the architect and what his / its intentions are.
mm you got a point!!

On a sunny day (Tue, 06 Dec 2022 12:26:20) it happened Wanderer@noemail.com
wrote in <8349@noemail.com>:

>On 12/6/2022 12:05 AM, Jan Panteltje wrote:
>> No Linux? No problem. Just get AI to hallucinate it for you
>> https://arstechnica.com/information-technology/2022/12/openais-new-chatbot-can-hallucinate-a-linux-shell-or-calling-a-bbs/
>>
>> Worries me more than a bit..
>
>It should. Imagine an AI assigned to each of us. Tracking us. Monitoring our behaviour. Knowing what we buy. What advertising we
>see. What we post in forums, product reviews, and emails. And then using that information to manipulate us. An AI that chooses
>what advertising we see. An AI that chooses what news headlines we see. What entertainment were directed to. How much of
>advertising is about the product and how much just manipulation? Pulling the strings of self image, desires, weaknesses for gambling,
>etc. How much is news and entertainment pushing an agenda? What if an AI could direct us to the ones that are written the right
>way to manipulate us. Pulling our strings. Turning us into puppets. Stealing our souls. And Google, Apple, Microsoft, etc. can
>say, "It's all encrypted. No person ever sees it. It's completely private."
>
>Of course, I'm too smart for it to work on me.

Musk is working on chip implants, those already exists from other companies,
So 'they' (Big Brothel) will require every newborn to have a chip implanted
that has some RF communication way and holds your credits, a killer serum (in
case you violated some thing) and you non-contact pay with it of course,
Maybe it generates a pain signal if you violate a traffic signal
or something...

Yesterday I was, for the second time, watching a TV program about anti-satellite
weapons (US space force etc), it showed how the first high altitude nuclear test
killed satellites (Telstar to be specific),

So.. bit of EMP and all of society including those with chip-implants and
all the AI is no more.
Mice and rats remain ??

On a sunny day (Tue, 06 Dec 2022 19:24:51 +0100) it happened Jeroen Belleman
<jeroen@nospam.please> wrote in <tmo1dj$1s7u$1@gioia.aioe.org>:

I kill-filter over 90% these days.

Be careful, you may have killed the God particle.

On Tuesday, December 6, 2022 at 8:30:02 AM UTC-8, bitrex wrote:

> My intuition is that the development of consciousness requires agency; a
> universe of some type to interact with in a meaningful way.
> Consciousness without agency sounds like a type of hell to me, ...

One organism that encouters this problem is botryllus schlosseri

<https://scopeblog.stanford.edu/2013/08/01/ocean-organism-settles-down-digests-its-spinal-cord-and-loses-its-individuality/>

bitrex wrote:
> On 12/6/2022 12:05 AM, Jan Panteltje wrote:
>> No Linux? No problem. Just get AI to hallucinate it for you
>>
>> https://arstechnica.com/information-technology/2022/12/openais-new-chatbot-can-hallucinate-a-linux-shell-or-calling-a-bbs/
>>
>>
>> Worries me more than a bit..
>
>
> "Degrave found that the simulation goes surprisingly deep."
>
> The mistake is believing it's actually simulating anything, it isn't.
> There isn't any logically consistent virtual machine operating under the
> hood, so commands of sufficient complexity will surely produce nonsense
> eventually.

People seem to have completely forgotten John Searle. Remarkable.

--
Les Cargill