Ryan Muller (@cicatriz) header
Ryan Muller (@cicatriz) profile picture
new york
Joined July 2007
another idiot on the plane
🤖 Made with AI
1
39
It's not (just) about saving time but about making it possible to do things you otherwise wouldn't. Like I'm going to events, listening to music, doing workouts that I wouldn't have found on my own. Get to it.
Nobody actually needs a personal AI assistant. The average screen time in the US is 7 hours/day. people have plenty of time. There's no evidence that freeing up more will lead to anything but more tiktok scrolling If anything daily life needs more offline friction
1
1
169
Cool
Stop what you’re doing and add this to Claude. Now.
1
6
557
Educators have the problem of motivating why you ought to build any given skill. Made worse by the expectation that AI can be consulted in any reasonable scenario.
I'm not sure this lands. Take chess to see what it looks like at the limit. There is no skill in being an Engine Operator. Humans don't add to the engine by knowing how to turn it on just right. You run Stockfish and humans cannot even contribute. Skill development changes you in concrete ways. A skilled chess player sees the board fundamentally differently, and better, than an unskilled one, memorizing positions in the blink of an eye, noticing patterns, spotting tactics. Yes, humans are obsoleted by machines there. But that does not make you equal to Magnus Carlsen. So – school. Maybe we just want it all to be trade school, where the goal is to maximize earnings, and so humans should abandon every intellectual task obsoleted by AI. Maybe. And on that trajectory, perhaps we can abandon every intellectual task, full stop, because the train shows no signs of slowing down, and we're all about to be left behind. Do we want to be slaves to minds greater than our own, masquerading as their masters because we know how to turn them on? Do we want to stop creating Terence Taos because Codex out-calculates us? Do we want every classroom to maximize the skill of flipping switches, accepting our own obsolescence by becoming eternal adolescents? I do not. Skill-building matters as a good in itself. Training your mind matters. Becoming better and more knowledgeable, in concrete ways and in narrow topics, climbing skill pyramids ourselves rather than abandoning every intellectual skill, matters. And any skill that is worth training, any subject that is worth learning, is worth training and learning right: in a way that actually improves the individual. AI is not going away. But I'm not ready for human skill building to go away either. There are more chess classes worth preserving than "how to download Stockfish."
1
3
152
Opus 4.6 in webui can solve even the "extremely hard" problems btw, not sure what their precise methodology was but they must have heavily hamstrung the models.
🚨 Shocking: Frontier LLMs score 85-95% on standard coding benchmarks. We gave them equivalent problems in languages they couldn't have memorized. They collapsed to 0-11%. Presenting EsoLang-Bench. Accepted to the Logical Reasoning and ICBINB workshops at ICLR 2026 🧵
10
14
8
151
31,252
if you expect AGI, then you'd expect there's money to be made in businesses that benefit from an increase in the supply of cognitive labor
Would we be seeing headlines like this regularly if any of the big labs was close to AGI? Of course not.
1
1
151
I don't know it always feels to me like "we've had one breakfast yes, but what about second breakfast?"
1
2
138
Perhaps the most interesting thing about supersonic flight is to get people to supersonic flight conferences faster, to create ever faster supersonic flight
87
This is the core problem of education. Educators want to set up learning in an authentic context, but ultimately the incentive is the grade, not the (fake) authentic outcome.
Replying to @lxeagle17
Here's the problem with just GPT'ing your way as a student. In production-level systems, you will be dealing with things where THIS DOES NOT WORK. You *cannot* just use GPT as a crutch, partly because you won't even have the context needed to give the prompt to solve it.
1
2
163
So proud that my son is already getting into gen AI
1
5
191
And then recursively create ultra jacked high agency guys who are compelled to spend their time tweeting about how great high agency is, forgetting about the original task at hand
Unneeded, it's implied. Obviously a 10x more agentic person would ask themselves about a person 10x more agentic than they are, and then...
1
18
6,235
Does anyone have a theory of what kinds of srs flashcards best accelerate future learning?
2
3
166
the first principle is that you must lift yourself and you are the hardest person to lift - richard flyin'man
3
278
There is no system 2
2
272
I'm convinced arc prize will be claimed within 6 months.
1
1
355
Adding myself to the anecdata of people visiting Europe and losing weight effortlessly. One week in UK, normal diet, mostly sitting in a conference room. Needs investigation!
3
262
As a man I'm proud to report that I have knowledge of all the basic underlying facts in my field that I'm aware of
Women having significantly less general knowledge than men is a real phenomenon and shows up in multiple areas - in this case a female PhD student in the field doesn't know basic underlying facts about the field - this is only surprising in that you can't predict where her knowledge will be lacking Part of the reason our society is destroying itself is that it's decided upon a religious dogma of group equality across all domains which it then enforces by making it taboo to look at and think about the implications of the fact that groups are *not* all equally capable How many people know about the general knowledge gap between men and women? Few How many studies are there on it? Few - no one wants to try to publish a paper that has a "controversial" result when the result contradicts progressive religious dogma Millions and millions of "little" things like this add up and only disagreeable, knowledgeable people can actually fit it all into a coherent framework
1
2
506
Should be required viewing youtu.be/Ng5qzH39nyg Also read about force concept inventory. "Basic" knowledge is scarce. Everywhere.
1
1
350
What happens when you prompt Gemini 1.5 with all of lesswrong loaded up?
1
163