Re: <nettime> The Baudrillardian Superintelligence Paradox: Capital's Terminal Simulation

[Pit Schultz via nettime-l <nettime-l@lists.nettime.org>]

Yann LeCun, speaking recently at the Grand Palais with its cathedral-like
acoustic reverb (still unmanaged by AI audio processing plugins), mentioned
that today an average house cat has more real-world (one-shot) intelligence
than the best-performing LLM.

For some reason, he recently switched from dogs to cats in trying to
allegorically demonstrate that the architecture of current large AI
projects aiming to "achieve AGI" is fundamentally flawed.

As a dog owner, I know how hard it is to train a dog, but also how easy it
becomes when you try to go with the flow and understand the motives or
embedded characteristics of your dog.

AIs are not far from that stage, even beyond the weak alignment "potty
training" to make them behave better. They're
philosophically/phenomenologically fundamentally flawed, but still quite
capable.

Word processors and typewriters had their impact on writing, and genAI will
have its impact too.

I recommend trying out NotebookLM, domesticated by writer Steven Johnson in
such a way that everything you input ends up in a specific kind of
patronizing left-mid-liberal inclined podcast dialogue between two
artificial radio hosts.[1]


The method I used here might be of interest, using a team of LLMs and then
finally letting the two with the best "style" and "character" condense the
final result in a battle mode.

I used Mistral, Llama, ChatGPT-Omni, Perplexity Sonar, Gemini-Flash,
DeepSeek-R1, and Anthropic Claude Sonnet in an iterative group think,
moderating them as hard as possible.

In the "collaborative" process, most of them clearly indicated that they
had been trained with openly available nettime data (without permission).

In this way, letting AI write the text takes longer than writing it
yourself. It might address the points you intended, but it will certainly
add new points and cut away others, only reproducing your own style of
thinking from a meta perspective.

By the way, for future parsing, it can help to disclose that AI was used.

Nevertheless, there are speech patterns so characteristic that it's still
quite easy to identify the writing style.

What they call AI slop today is probably the next retro charm of media
massage.


[1]
https://notebooklm.google.com/notebook/c7862110-2b14-4a97-bea1-e7a3191d437e/audio


p.s.
Btw, it´s funny, when former post-structuralists and far-leftists discover
that their late sympathy for humanist German idealism, or rather the
romantic Biedermeier version of it (unheimlichkeit), takes a Spenglerian
turn. it's even funnier that it's not the critique of content but the form,
when the construction of authorship turns into an essentialist argument for
auratic originality. i'd rather try to find counter-narratives to
deterministic thermodynamic "scaling laws", TESREAL aka californian
ideology 3.0, when even the war room of wired magazine is almost ready to
escape from san francisco. didn't foucault dream of the numerability of
fundamental utterances in the archive as a sign of the technology of power?
if large amounts of compute are alienating the already instrumentalized
rest of us, why not.


On Thu, Feb 13, 2025 at 4:01 PM Frédéric Neyrat via nettime-l <
nettime-l@lists.nettime.org> wrote:

> hi sh:
>
> excellent question! why using a camera, and which one? If questioning the
> technology we use, when we use it and why, is meaningless, it confirms
> Bifo's point about AI & dementia.
>
> best,
>
> fn
>
>
> On Thu, Feb 13, 2025 at 8:11 AM Stefan Heidenreich via nettime-l <
> nettime-l@lists.nettime.org> wrote:
>
> > Hi,
> >
> > > Why did you decide to use AI to generate this text?
> >
> > isn't that a funny question? Soon it will sound like in the 19 century:
> > 'why did you use a camera to make that image?' or 'Images/texts
> > generated by camera/AI or not real art/thoughts.'
> >
> > An btw: I guess it's a pun anyway. How long did it take you to generate
> > the msg you like. How much time did you spend to adjust the prompt (the
> > camera)?
> >
> > best
> > sh
> >
> >
> >   Why this decision, what
> > > is its meaning, its purpose? You can use AI to answer my question,
> which
> > > would be an answer as such (a tautology actually, a mediated answer
> that
> > > would confirm what sort of message it is, to borrow from McLuhan). If
> you
> > > answer my question with the help of any AI, I wonder how far this
> > > decision should, retroactively, question your first post and change the
> > way
> > > to read it.
> > >
> > > Best,
> > >
> > > Frédéric
> >
> > >   (LLL, 2025)
> > > __________________________________
> > >
> > >
> > >
> > >
> > > On Wed, Feb 12, 2025 at 4:37 PM Pit Schultz via nettime-l <
> > > nettime-l@lists.nettime.org> wrote:
> > >
> > >> The Baudrillardian Superintelligence Paradox: Capital's Terminal
> > Simulation
> > >>
> > >> Sam Altman's three scaling laws for artificial intelligence -
> > logarithmic
> > >> intelligence gains, hyper-deflationary costs, and super-exponential
> > value -
> > >> mask capitalism's terminal phase: an accelerated collapse into
> > algorithmic
> > >> hyperreality where AI-generated market simulations supersede and
> > ultimately
> > >> consume material reality. A Marxist-Baudrillardian synthesis allows us
> > to
> > >> map how superintelligence triggers financial implosion. This occurs
> > through
> > >> three interlocking mechanisms:
>
>
-- 
# distributed via <nettime>: no commercial use without permission
# <nettime> is a moderated mailing list for net criticism,
# collaborative text filtering and cultural politics of the nets
# more info: https://www.nettime.org
# contact: nettime-l-owner@lists.nettime.org