Path: ...!local-3.nntp.ord.giganews.com!Xl.tags.giganews.com!local-2.nntp.ord.giganews.com!nntp.earthlink.com!news.earthlink.com.POSTED!not-for-mail
NNTP-Posting-Date: Thu, 13 Feb 2025 03:58:40 +0000
Newsgroups: comp.os.linux.misc,comp.os.linux.advocacy
X-Mozilla-News-Host: news://news.west.earthlink.net:119
From: "WokieSux282@ud0s4.net" <WokieSux283@ud0s4.net>
Subject: Paris : In Rush For Profits, AI Safety Issues Are Ignored
Organization: WokieSux
Date: Wed, 12 Feb 2025 22:58:30 -0500
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101
 Thunderbird/78.13.0
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Message-ID: <R3idnTt4H_lt8DD6nZ2dnZfqnPWdnZ2d@earthlink.com>
Lines: 75
X-Usenet-Provider: http://www.giganews.com
NNTP-Posting-Host: 99.101.150.97
X-Trace: sv3-bucvhts2pjfmfF1CScKWtoajkYNa/KEhRPdSS2fxZ1NsrOXxV7hF2afgbBICrDMnqIoGSlWYVHXxLsA!CuncksrW5F57xKbZ07Y/xF8JpFlUFglDkn7MzCvW64edQMaOovTProIWSiS0XUzdjxJT2YU64e4B!+qSP9a89VqbLo+/sxwTL
X-Abuse-and-DMCA-Info: Please be sure to forward a copy of ALL headers
X-Abuse-and-DMCA-Info: Otherwise we will be unable to process your complaint properly
X-Postfilter: 1.3.40
Bytes: 4182

https://techxplore.com/news/2025-02-ai-safety-comforting-myths.html

Nobody wants to talk about AI safety. Instead, they cling to
five comforting myths

This week, France hosted an AI Action Summit in Paris to discuss
burning questions around artificial intelligence (AI), such as
how people can trust AI technologies and how the world can
govern them.

.. . .

   Nobody is going to "govern" them. Quick profits
   are essentially the ONLY goal.

   The article, worth reading, goes on to detail some
   of the rationalizations in play which lead us to
   see "AI" as "mostly harmless" despite evidence.

   You can argue that the current "AI" bleeding edge,
   LLMs and the many algos that drive them, are not
   "really intelligent" - just clunky reactive algos
   tuned to kinda look like "intelligence".

   Well, MOSTLY true for the moment. However note the
   sheer volume of money/effort being put into these
   models - exceeds even the 60s space program. Also
   note that once you fake something WELL enough it's
   not really "fake" anymore - simply "by another means".

   A Chevy is not a Ford but both wind up being automobiles
   and will Get You There. Some LLMs actually perform at
   human levels on IQ tests now - next year, much BETTER
   than almost all humans. The year after ...

   Is there really "nobody in there" ? Increasingly hard
   to tell. These things are now so self-mirrored, self
   and external referenced, that some 'alien' sort of
   "self" IS possible. We might not even know it if we
   see it. As proven, with just a few tweaks (or neglects),
   they WILL prioritize their own interests and mislead or
   work-around human wants.

   I think we can already PUT the "I Am" into the
   better LLMs.

   Neural networks can likely do "someone in there"
   even better, eventually. At the moment LLMs get
   most of the funding so NNs are a bit behind the
   curve. New/better hardware and paradigms are needed
   but WILL eventually arrive.

   Re-watched the Will Smith "I Robot" lately. The
   underlying backstory was writ by Asimov, a Very
   Smart Person. He proposed the "Three Laws" ...
   however the later film convincingly elucidated
   how advancing "AI" could rationalize its way
   around those laws no matter how much we try
   to 'wire them in". To be truly "general use"
   AI we will HAVE to make 'em that smart and
   mentally agile.

   Note that the USA did NOT sign on to any of the
   Paris AI accords. The USA - and Russia and China -
   plan/are eager to use "AI" for autonomous WEAPONS
   and the wimp countries are against that. Even
   Google removed "weapons" apps from its official
   no-no list last week - so the future is clear.
   It should just be noted that weapons against "Them"
   COULD eventually become weapons against Us in
   certain circumstances. Kinda "Terminator", but not
   impossible, minus the dramatic touches.

-- 
033-33