US Govt Pols Now Backing 'AI' "Pause" Recommended by Musk, Wozniak

[22T.R732]

The US govt is getting involved in 'AI' development.

As top tech people have become spooked by ChatGPT and
similar software they urged the govt to take action
to "evaluate" the downstream effects of Chat on all
manner of human activities, including job markets.

Chat 3.5 is already scary "smart" - it can pass the
old Turing Test easily ... coming across as some guy
you met at Starbucks. Chat 4.0 is almost ready for
prime-time ... wider capabilities, smoother delivery.

Chat also has the unsettling ability to write, and
then improve, computer programs to handle whatever
you want. This is basically the ability to rewire
its own "brain".

What would happen if you let a bunch of Chat instances
talk to each other for awhile ?

Rule of thumb - if you fake something WELL ENOUGH
then it isn't fake anymore.

In any case, govt pre-emptive suppression of a rapidly
emerging broad technology based on worrisome possible
uses/misuses is rather rare. Likely some EU govts will
take the same tact.

Of course Russia and China have likely already stolen
the Chat code and will put it to THEIR uses ... which
will quickly result in an "AI Gap" with serious national
security implications.

Seems like someone in comp.ai deleted the earlier version
of this - so I'm spreading-out this one so you can't get
at every copy. Gonna SAVE it too ... and have no problem
re-posting every day for months. What Musk and friends
have done directly impacts all 'AI' work - commercial
or academic. You did your jobs TOO well - now it's
frightening people - and Chat (so far) doesn't even have
an actual 'consciousness'.

[Tristan Miller]

Greetings.

On 2023-03-31 06:31, 22T.R732 wrote:
> Seems like someone in comp.ai deleted the earlier version
> of this

Your earlier post was automatically rejected by the moderation software
for excessive cross-posting. (You had submitted your article to
talk.politics.misc, alt.survival, alt.politics, and alt.politics.usa in
addition to comp.ai.) Don't cross-post and your submissions will not be
automatically rejected.

Regards,
Tristan
comp.ai moderator

--
Dr.-Ing. Tristan Miller, Research Scientist
Austrian Research Institute for Artificial Intelligence (OFAI)
Freyung 6/6, 1010 Vienna, Austria | Tel: +43 1 5336112 12
https://logological.org/ | https://punderstanding.ofai.at/

[22T.R732]

On 3/31/23 3:12 AM, Tristan Miller wrote:
> Greetings.
>
> On 2023-03-31 06:31, 22T.R732 wrote:
>> Seems like someone in comp.ai deleted the earlier version
>> of this
>
>
> Your earlier post was automatically rejected by the moderation software
> for excessive cross-posting.

I am allowed 5 x-posts - so I used them.

The subject in question touches on a number
of relevant groups. This is not just an
"academic" exercise anymore - but a social/
political thing at the highest levels.

Amazing how it all moved so quickly.

In comp.ai there is a very relevant issue as to
how this will affect the more "academic" aspects
of 'AI' research, short/medium-term. There is also
the relevant, 'Frankensteinian', aspect of having
to consider what the fruits of genius might actually
do to the world.

What I *objected* to was that your deletion HERE
managed to propagate to all the OTHER groups included.

But nevermind, I seem to be getting fair traction
elsewhere. This HAS become one of the "major issues
of our time" it seems. Won't be long at all before
the US govt, likely the EU as well, starts making
edicts (probably ill-considered). Musk, Woz and major
friends have now STARTED something. I think their
letter may reach similar importance to the one
Einstein and friends sent to Roosevelt about the
possibility of atomic weapons.

> (You had submitted your article to
> talk.politics.misc, alt.survival, alt.politics, and alt.politics.usa in
> addition to comp.ai.)  Don't cross-post and your submissions will not be
> automatically rejected.
>
> Regards,
> Tristan
> comp.ai moderator
>

I *think*, long long ago, I had conversations with
Marv Minsky in this group.

[Tristan Miller]

Greetings.

On 2023-04-02 14:07, 22T.R732 wrote:
>> On 2023-03-31 06:31, 22T.R732 wrote:
>>> Seems like someone in comp.ai deleted the earlier version
>>> of this
>>
>> Your earlier post was automatically rejected by the moderation
>> software for excessive cross-posting.
>
>   I am allowed 5 x-posts

Says who? Every moderated group (and indeed, every news client and
every news server) can set its own limit on the number of cross-posts.
You need to comply with *all* of these limits in order for your posts to
make it through.

>   What I *objected* to was that your deletion HERE
>   managed to propagate to all the OTHER groups included.

I didn't delete anything, and so there was nothing to "propagate". Any
time you submit an article to a moderated group, even if it is
cross-posted to other unmoderated groups, then it needs to pass
moderation *before* it appears anywhere on Usenet. Your article was not
deleted from the other groups because it never appeared in them in the
first place.

Your article was clearly on-topic here and you're welcome to continue
the discussion here so long as you don't cross-post. I've started
another thread on the question of whether to allow cross-posting to comp.ai.

[26B.X919]

On the whole I agree with your concerns. I started with Usenet
way on back, when it was still "civil". It is no longer "civil".
Allowing cross-posts, or many cross-posts, is likely to result
in chaos, flame-wars and worse. I have seen many groups go that
way and a more 'academic' group like comp.ai does require some
sort of shield ... yet one that cannot over-tax any moderators.

However 'AI' has finally reached a state and form where it is
no longer just "sci-fi musings" but a tangible, sometimes
threatening, influence on global civilization. Musk and a fair
number of other prominent tekkie figures are now seeing that,
and SAYING something about it. This has attracted considerable
attention, even at the highest levels - regulatory/policy/funding.

As such, 'comp.ai' is no longer as obscure as it ought to be,
it's suddenly part of a big 'something'.

I am not sure what is required these days to create a new
Usenet group. I would suggest something like "comp.ai.policy"
to absorb the large-scale, oft crap, postings while still
being sort of "on target" to those most involved in 'AI'
research.

The near-term future of 'Chat', 'Bard' and related approaches
to, hmm, 'human emulation', is not entirely clear. There DOES
seem to be a potential for rapid SELF-improvement with these
sorts of systems. The human mind is in many ways a "hall of
mirrors" and that effect CAN be emulated ... the system
observing itself, the world, the effects of the world on
itself and so forth - making continual improvements.

People, politicians, money interests and such DO need to worry
about a Chat-5.0. There are likely many paths to de-facto
"consciousness" and Chat and/or friends may indeed be one of
those. "Emulate", "fake", something WELL enough and it is not
really "fake" anymore ... just "consciousness", "being", by
alternate means. That has broad *implications* that have,
until now, been pure sci-fi.

I saw in the news yesterday that Wal-Mart corp is going to
fire most of its human warehouse/shipping humans and
replace them all with 'AI'/"Expert Systems". It tried to
spin that as the fewer humans getting easier, better-paying,
jobs. However that leaves the 99% unemployed. Not the ONLY
corp moving in that direction. This is serious, this is of
economic significance, this is *political*.

Remember about a year ago when a Google 'AI' expert came
to the conclusion that his system WAS effectively
'conscious' ? They FIRED him for saying so.

Hmmm ... how many 'AI' profs/experts can Chat-5.0 replace ?

Beginning to 'get it' ??? :-)

Mary Shelly was prescient.