Simon @spoltier@qoto.org
- Country
-
- profile banner
- https://showyourstripes.info/
code / data wrangler in Switzerland.
Recovering reply guy. Posts random photos once in a while.
Joined Jul 2023
Simon
boosted
@shironeko @andrewrk @cadey thanks, I would have walked right into that
Simon
boosted
Working on a draft. Is it too hard to lead with this?
The deployment of generative artificial intelligence tools has been a disaster for the human race. They have allowed a select few to gain "higher productivity"; but they have destabilized society, have made work transactional, have subjected artists to indignities, have lead to widespread psychological suffering for the hackers that build the tools AI companies rely on, and inflict severe damage on the natural world. The continued development of this technology will worsen this situation. It will certainly subject human beings to greater indignities and inflict greater damage on the natural world, it will probably lead to greater social disruption and psychological suffering, and it may lead to increased physical suffering even in "advanced" countries.
Simon
boosted
It's another kind of space race, so we all "win" in the end. https://blog.codinghorror.com/the-2030-self-driving-car-bet/
Simon
boosted
Some notes on Shisa v2 405B, a new open weights LLM release from Japan that's an example of Sovereign AI - the ability for nations to build models that reflect their own language and culture
https://simonwillison.net/2025/Jun/3/shisa-v2/
@codinghorror I think you will win but fairly narrowly - e. g. Chicago doesn't fulfill the criteria because of winter weather, or there are still remote operators (the strict interpretation of level 5 would exclude those I think)
I assume you want attract attention to the progress in America specifically, but I will note that there are probably more driverless rides taken, in more cities, in China; this is more a question of regulations and the will to implement than basic science / technology.
Simon
boosted
Simon
boosted
Last week, @steveklabnik.com indicated his disappointment with AI discourse:
https://steveklabnik.com/writing/i-am-disappointed-in-the-ai-discourse/
Perhaps unsurprisingly, @ahl and I are going to be joined by Steve today to talk AI discourse; join us, 5p Pacific!
Simon
boosted
An extra special Cat Talks announcement: I'm going to be at Comic-Con to join a panel exploring the latest season of ANDOR: "From Resistance to Rebellion." I'll be repping as a psychological scientist sharing about the psychology of coalitions, the psychology of conformity, and the psychology of courage under authoritarianism!!!!!!!!
Details to follow closer to. But yes, this IS the most Dr. Cat event imaginable 
Simon
boosted
Was reminded of this great piece today by another tech dopamine post
https://medium.com/the-spike/the-crimes-against-dopamine-b82b082d5f3d
Simon
boosted
Some notes on the new DeepSeek-R1-0528 - a completely different model from the R1 they released in January, despite having a very similar name
Terrible LLM naming has managed to infect the Chinese AI labs too
https://simonwillison.net/2025/May/31/deepseek-aideepseek-r1-0528/
Simon
boosted
This is likely a historic mistake, but I’m starting to wonder if the line between “science” and “engineering”—which we’ve learned to bracket in the case of CS—could get bracketed in certain social science contexts as well.
RE: https://bsky.app/profile/did:plc:565ebob5f6hw33hjdkxty6qj/post/3lqgjx4aqls2x
We can learn about the world by tinkering and changing it; we’re not a federation starship governed by the prime directive.
@tedunderwood.me
Could we say: If the practitioners are not clearly distinguishable from scientists, it's a sign that a field is preparadigmatic? See also medicine, and social sciences as you mention.
(I'm assuming you mean mainly ML for computer science. In other cases I would say that the distinction is more clear)
@nyhan @roadskater @hunterking @sunscream he also mentions that he open sourced the code.
The compliance code is 800 lines of python that extracts text from pdfs and asks an LLM to check for reference to relevant topics. Seems a bit like open-sourcing anti-cheating tools, but 
https://github.com/lecy/va-doge-tools/blob/35e3ff1b9e0eb1c8aaaebf3bfe76f2002354b782/eos/analyze_eos.py#L218
Simon
boosted
@roadskater @hunterking @sunscream apologize in case there is a response I'm not seeing (Mastodon does that), but I do have to point out that the OP linked fast company article starts with the following:
> Sahil Lavingia has had just three jobs over a 15-year career in tech.
So not a young idealist, in fact a tech multimillionaire; who was naive about how government works, which is probably the case of most tech multimillionaires.
Simon
boosted
Simon
boosted
How are we doing at "being technical"? @grimalkina - looks at how technical teams work, finds it fascinating!
Questions for engineers:
- what does your work help people do?
- how many people a day does your people help?
@nic221 I'm not sure if he's doing it on purpose to turn off the people who disagree, but he should learn to spell principle correctly. That's the one thing I have no nuanced opinions about: _decelerate_ not giving a shit about spelling.
Simon
boosted
A small slide deck for a 15 minute impulse talk at Cycon 2025 in Talinn: https://docs.google.com/presentation/d/1_3Iu74UijAjfSLHzqWDkDEaIwoB6WBSo9-mY5e0u0HM/edit?usp=drivesdk
- Country
-
- profile banner
- https://showyourstripes.info/
code / data wrangler in Switzerland.
Recovering reply guy. Posts random photos once in a while.
Joined Jul 2023
