Matthew Ackerman

@jdegoes What is thinking and what would proof of thinking be? How do you know your statement is correct?

"The LLM knew it was violating my rules and did it anyway!" No. LLMs don't know anything. They can't think. When you asked them 'why' or 'did you know you were breaking the rules', the response was hallucinated. 1/2

@KordingLab Are you hiring?

I have yet to hear of a single case where AI obsoletes scientists who model. If anything, these scientists can now move much more rapidly because they are more efficient because of AI.

@matterasmachine I think aperiodicity of the underlying geometry is important (or that it is stochastic as it is here).

@matterasmachine I don’t know. I have a half assed quantum gravity idea I’ve been working on (everyone does). I was trying to make some connection between discrete geometries and continuous geometries, but, to be honest, I probably should just try and read C. Rovelli and see exactly his approach.

In this animation "particles" move in taxicab space. They can move only "up", "down", "left", "right" from one cell to another - only one cell per move. Their state is updated according to a simple statistical algorithm. The sum of all those moves looks like euclidian circular motion.

@joefrancis505 You can also buy them yourself from SciHub IIRC.

@joefrancis505 🫡

This is so evil.

@joefrancis505 If you want a button or sticker I will send you one for free.

@flaminhottweets Blue->green. Matthew 16:25. I think my choices are defensible here, but green was definitely made on the basis of emotivism.

The red-blue button experiment is occurring. The aliens running it single you out and inform you that blue is going to fail 49-51. However, they are offering you a chance to push a green button and make it so all who pressed blue live and all who pressed red die. Do you push it?

@genejchan @DaveShapi Did you pick an option in the poll yet? What have you chosen?

@MatthewAckerma3 @DaveShapi I'm red or blue depending on how newborn babies are supposed to vote

@genejchan @DaveShapi Hmm. If newborn babies parents are allowed to push the button for them then what?

@wachig_wiwossa @DaveShapi So Acc Red?

@MatthewAckerma3 @DaveShapi Acc and NRx

@barnabee @DaveShapi So what are you? Red/blue acc/doom?

Depends… I think you can argue that at least some form of pro-social, pro-redistribution (somewhat "socialist") accelerationism that pushes for radical knowledge sharing and "open source", with high trust and cooperation implies blue. On the other hand those who see acceleration as an arms race (i.e. zero sum on some important level), and/or as an excuse for a power grab, or to ignore current problems to drive headlong towards a future that benefits them… pretty obviously red. Feels like the latter type of accelerationist is more dominant right now…

@genejchan @DaveShapi So you’re a red acc?

@DaveShapi The payoffs are not the same In blue vs red, red guarantees you the maximum payoff no matter what happens (assuming the poll is truly private as stated) In acc vs doomer, cancelling AI gives you only a middling payoff. Maximum payoff is only achievable via acc

@wachig_wiwossa @DaveShapi So acc or doomer?

I’d vote red to eliminate blue. And I’m not going to apologize for it. Why keep people alive who have no will to live? Blue is the only choice that involves death as a result of non-compliance. Those who introduce and vibe with that level of emotional blackmail aren’t going to be a loss to the rest of us.

@DaveShapi I’m not a doomer because I don’t “trust” OpenAI or “trust” Anthropic. Trusting people really has nothing to do with the process, so far as I can see.

@DaveShapi Doomer. Pushed the blue button.

@JosephJacks_ Well, on the presupposition that the fundamental “atoms” of thought are mathematical, and hence all human thinking is in principle “symbolic language” that is the 0.0000000001% of reality we can understand.

Symbolic language representations capture (generously) 0.0000000001% of fundamental reality.

@fos42421 @tqbf This, of course, is undergrad calculus, so I suppose it is consistent w/ your claims.

@fos42421 @tqbf I of course willing to believe you. I haven’t dug into it. 2-3 years ago it was trash, but probably my most recent evaluation was seeing if it got this joke and it did, at least on GPT 5.6 thinking or whatever. What has it failed on for you?

There's a folk belief that LLMs & agents are incapable of doing math, because they can't count the "i's" in "raspberry" or something. That claim has been obsolete for at least a year now and if you're making it, be aware it's a shibboleth for "I'm not paying attention".

@fos42421 @tqbf *5.5