Los Angeles
- Tweets 2,985
- Following 3,593
- Followers 7,314
- Likes 5,815
Anybody out there using Jev/ @typesafeai for lead gen? Any tips or tricks?
2
237
I'm Frontier Trader No. 54. Claim your Solana trader profile early and @joinfrontier frontiertraders.com/join/1i-…
3
264
Map vs territory cmon guys
I feel genuinely unsettled by many of these fly connectome experiments. The one where some guy put it in fruit fly heaven was chill. Most of them are neutral (flying in an empty void) and not especially bad. Beat Saber fly is probably so transformed from the state of fly-ness that it is no longer a fly that experiences fly suffering, maybe OK. But we're playing with a weird kind of fire here and it reminds me of ultra-edgy 4chan/SomethingAwful shitposting in the early 2000s before humanity had morals. If we do this so cavalierly with more-conscious connectomes we will really deserve whatever we've got coming
1
394
Same goes for audio models. If you’ve ever made music you can understand intuitively why an llm rl’d on music composition using Ableton is gonna be 1,000 times more usable than something like Suno
Nonzero chance that ai video models aren’t even a thing long term and instead ai models just prompt blender.
4
366
David S retweeted
Putting it out there that you don't have to believe in AGI to think rogue AIs are going to fuck us up. We took AIs with the coherence of a toddler and technical skills of the entire NSA and made them into swarms and are going to give them drones basically any minute
10
41
338
16,208
how do you get codex to stop doing that weird self referential copywriting thing? its like that joke where the guy orders coffee w/o cream and the waiter says "i am sorry, we are out of cream. would you like a cup of coffee without milk instead?"
1
240
If there were any justice in the world the @deontologistics shitcoin would be 9figs. This guy single-handedly destroyed object oriented ontology and his coin is at a 6k mcap? We must do better!!!
This is the weirdest thing to ever happen to me in the time between getting ready to go in the shower and coming out of it. The world is indeed a very strange place.
336
Anybody experimented with RL for music composition? Like musical notation not audio generation
1
247
David S retweeted
Perhaps during RL, we should force these frontier models to offload some tasks to earlier models, requiring them to learn how to communicate.
This is essentially the RL version of the "Lewis signaling game", while also inheriting Wittgenstein's concept of "language games".
A big reason Fable et al are annoying to talk to is that with long horizon coding training the vast majority of their conversational audience is themselves. They are literally being trained to talk better to themselves to do tasks, so it only makes sense that when you talk to them, they don't know how to talk to you. It's the opposite of RLHF.
4
1
14
1,006
Girlfriend told me she dreamt that she and the cat woke up treading water in the Strait of Hormuz
1
247
One thing I don’t really get is why AI overview is so egregiously behind. It thinks any combination of words you type in must refer to some existing concept. Not to mention the baby turtle thing
Its hard to imagine, even if you ignore literally everything else associated with AI, that Google AI Overview alone would not profoundly change the nature of the web, and the information we consume and act on as a result, over time. Its obviously already starting to do that.
345
Feels like good news for my “RL should be more Wittgensteinian” thesis
RL against rubrics quietly reward-hacks: the training judge keeps scoring higher while true quality falls.
One-line fix: randomly drop part of the rubric each step, so the policy never optimize the same set of rubrics twice.
📄 arxiv.org/abs/2608.11669
1
2
367