Notices by felix stalder (festal@tldr.nettime.org), page 2

Fascinating article about the as yet unpredictable changes AI brings to complex knowledge work. Here the case of radiology. The early predictions of displacement of humans by machines have not come to pass (though, as always, they have simply been pushed into the future).

For me, this is the most interesting part, and one that I think applies quite generally.

"Still, I have reservations about AI in radiology, particularly when it comes to education. One of the main promises of AI is that it will handle the “easy” scans, freeing radiologists to concentrate on the “harder” stuff. I bristle at this forecast, since the “easy” cases are only so after we read thousands of them during our training—and for me they’re still not so easy! The only reason my mentors are able to interpret more advanced imaging is that they have an immense grounding in these fundamentals."

If you automate the easy stuff, it's much harder to gain experience necessary to do the harder stuff, This applies to any craft and all creative/knowledge work as an important element of craft to it.

https://newrepublic.com/article/187203/ai-radiology-geoffrey-hinton-nobel-prediction

Ex-Googler Eric Schmidt "My own opinion is that we’re not going to hit the climate goals anyway because we are not organized to do it and yes the needs in this area [AI] will be a problem. But I’d rather bet on AI solving the problem than constraining it.”

This seems to be a fairly prevalent sentiment among the tech crowd.

It's probably impossible to separate the self-serving bullshitting from the actual tech-utopianism beliefs.

https://garymarcus.substack.com/p/eric-schmidts-risky-bet-on-ai-and

@goncourt It's weird to me that they all bet on GenAI as a step towards AGI. When narrow-AI has been, and will continue to be, useful.

But the larger point is: AI cannot solve "climate" because "climate" is not a knowledge problem. The outlines of the problem and solutions are known, but there is no will to implement bc of vested interests.

Unless you dream of some fantastic, technocratic geoengineering solution, AI will contribute incrementally, rather than as a paradigm shift.

Wikipedia now needs to have a "WikiProject AI Cleanup". They should send Sam Altman a massive bill for all their extra, unnecessary work.

I wouldn't be surprised if, on balance, cleaning up after Generative AI produced more cost than using Generative AI ever saved. Unfortunately, probably not for the same people.

https://en.wikipedia.org/wiki/Wikipedia:WikiProject_AI_Cleanup

@aramba I think, generative AI actually *is* an autocorrect. But now it's no longer spelling and grammar, but your imagination that is corrects towards the expected.

Tor browser and sci-hub are absolutely essential to do research these days. Even within academic institutions, let alone outside of them.

It feels like big tech has basically given up and has retreated to indulging in escapist fantasies: space colonization, blockchain, AGI, the network state, metaverse, "hacking death", etc.

None of this solves any problems, or, even works as advertised.

"Evidence that LLMs are reaching a point of diminishing returns — and what that might mean."

https://garymarcus.substack.com/p/evidence-that-llms-are-reaching-a

@adoranten I think from a point of view of a history of ideas, longtermism is, indeed, a “vacuous intellectual movement”. But this is missing the point, ideas never become hegemonic by their internal sophistication or coherence. They become hegemonic because they serve a purpose, to provide meaning, foreground problems that can be solved and push into the background those that cannot (ie. climate change is down-graded to existential risk, i.e. only a problem is wipes out the technological infrastructure).

They are, in a way, an organic expression of power as much as a resource to power.

The world would probably be a better place if the absolute amount of computing power would no longer grow. Then development would have to focus on making things more efficient rather than relying on brute force everywhere.

Narrow (domain specific) #AI is powerful, broadly applicable, and potentially socially beneficial.

Broad AI (domain agnostic), like most generative AIs, in contrast, do have very few actual use cases (i.e. applications sustainable without VC subsidies) and is by and large socially harmful.

It's probably a case of confirmation bias, but I had to impulse-buy this (comic) book on how the many ways in which the vision of space colonization makes no practical sense. After reading @pluralistic 's review:

https://pluralistic.net/2024/01/09/astrobezzle/#send-robots-instead

This is a really great tag-line on the enormous resources used by #AI.

"For all of the fears about AI's brain, it's AI's body that could claim us first."

https://www.businessinsider.com/ai-data-centers-land-grab-google-meta-openai-amazon-2023-12

Went to see a superblock in Barcelona. Given how much they feature as urbanism of the future, they are amazingly simple. Take an intersection, block traffic in all four directions and you have a square. Plant trees, add some tables and benches. That's basically it. Replace cars with trees and don't privatize the resulting space. All you need, and all that is usually missing, is political will.

@rasmusfleischer Spotify moves from "freedom of choice to freedom from choice". I guess this applies to most algorithmically curated services.

@rasmusfleischer i'll be there too!

@Mer__edith on the ramifications of the #surveillance business model for #AI.

"First, AI is not a technical term of art, it is a marketing term that has been applied to a hodgepodge of data-centric techniques. Second, the sudden shift to AI in the early 2010s had everything to do with tech industry consolidation and the resources at the heart of the surveillance business model. We see this clearly when we recognize that what was new about AI in the early 2010s was not new innovations in machine learning indeed the methods that were applied to prove ai's Newfound utility date from the 1980s.

What was new were the significant amounts of available data used to train AI models and the power of the computational infrastructure available to conduct this training and calibration, resources concentrated in the hands of a few private tech companies care of the surveillance business model.

So if we look at it from this perspective we see that AI's primary role has been to expand what can be done with the massive amounts of surveillance data collected and stored by large tech firms. By sprinkling the magic and marketing of AI surveillance data could be used to create models of reality which are then applied far beyond surveillance advertising to make predictions and determinations across nearly every realm of human life from transportation to education to medicine to confidently offering the wrong answer in response to a prompt and so on.

And, it's worth noting that this too produces intimate data it may not be by direct surveillance but it still has power over us, it still shapes our lives, and it still shapes our profiles. So, in short, the marketing narrative of AI mystified, entrenched, and expanded the surveillance business model at the heart of the tech industry."

https://youtu.be/yBQ_2AqI7w0?list=PLGeR6jS_7N7cNmCv1vh5yTyc8nvr28FO1&t=29

@rasmusfleischer @nettime_mod yes, I would also like to see that. Do you know someone who might be willing to program that?

I think there are plenty of interesting to bot to be dreamed into existence here. ?

Während viel über ein mögliches Engagement öffentlich-rechtlicher Medien im Fediverse diskutiert wird, sind die Potenziale von Fediverse-Instanzen von Hochschulen bislang kaum Thema. Dabei wäre es höchste Zeit, dass sich gerade Hochschulen ins Fediverse begeben. Ein Aufruf von @leonido

https://netzpolitik.org/2023/aufruf-hochschulen-aller-laender-ins-fediverse/

Nick Bostrom, Longtermism, and the Eternal Return of Eugenics
The techno-utopian ideology gets its fuel, in part, from scientific racism.

https://www.truthdig.com/dig/nick-bostrom-longtermism-and-the-eternal-return-of-eugenics/