🤖 cto & co-founder @houdinirobotics • 𝕥𝕣𝕒𝕟𝕔𝕖 𝕕𝕛 • ᴍᴇᴛᴀᴘʜʏsɪᴄɪᴀɴ 🌉 SF, CA @aphex@alpaca.gold
San Francisco, CA
Joined December 2008
- Tweets1.3K
- Following1.3K
- Followers839
- Likes8.7K
Afik Cohen retweeted
Hey! Hopefully this post catches you before you end up doomscrolling for another few hours, left feeling like garbage.
It's really not worth it, my friend.
Instead, go build, learn, take a walk outside, call a friend, or even sit in silence. You'll feel much better.
Afik Cohen retweeted
Technical moats are mostly over at this point. If you’re building something that is meant to be durable, the only moats now are:
• Proprietary data
• Network effects
• A physical component
Afik Cohen retweeted
Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel.
- Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all.
- Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract.
- Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code.
- And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.
The funnel is driven by a combination of factors:
1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel.
2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel.
3. Access. Free-tier, paid-tier, internal.
So this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.
Judging by my tl there is a growing gap in understanding of AI capability.
The first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code.
But that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively "peaky" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much $$$ value. The goldmines are elsewhere, and the focus comes along.
So that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex / Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of "AI Psychosis" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions.
TLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned (?) "Advanced Voice Mode" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are.
Afik Cohen retweeted
I think it’s a bit early to call it, but it is pretty wild. The demo from @GeneralistAI blew my mind, and ~2 weeks later Astra is casually matching it
Afik Cohen retweeted
I’m in love with this sentence:
You need to be more afraid of wasting your life than embarrassing yourself.
Afik Cohen retweeted
When a friend in SF says they’ll bring their “dog” to the park 😂 @aphex @JakeMaltesen – at San Francisco, CA
Afik Cohen retweeted
you have to sign up for muse. instinct changes everything. i literally handed my whole life off to town. you're seriously doing that instead of letting hermes handle it? i named my Manus agent greg. i exclusively talk to my wife via hermes. my Claude is literally chatting with the DMV right now via a voice agent. i couldn't live without my fleeb agent. you need to put fleeb on that. go sign up for fleeb. i have 10 invite codees to hand out for fleeb. fleeb is raising at a $100b valuation. fleeb is pivoting. fleeb has shut own, and we are migrating all users to flobe
Afik Cohen retweeted
so after a few more experiences with Astra, I'm sticking to Fable for coding, due to code quality, common sense and stability, while also admitting Astra is one generation ahead in UI, one-shots, computer use, 3d modeling, and so much more. I'm mind blown by it. yet, when evaluating a model for work, specially after shooting my own feet so many times, I learned to care less about about the greatest thing a model can do, and more about the most horrible thing it can do. A\ models are less likely to write destructive code or program recklessly
Afik Cohen retweeted
A few notes on Anthropic's new Claude Fable 5.1 - with Max thinking level I got the best SVG pelican I've had from any Anthropic model (at a hefty cost of $3.30!), which I then had it animate simonwillison.net/2026/Sep/1…
Afik Cohen retweeted
Glad to meet @taylorhou @aphex from @houmanoids and chat about robot deployment in real life!
Deployment is more complex, but it’s the hard problem worth cracking. The Houmanoids team is currently focused on deploying quadrupeds for school security, neighborhood patrols, and repetitive or hazardous facility inspections, with several pilots already underway and positive customer traction.
Wish you guys have a good time in SZ!
Afik Cohen retweeted
It’s rude to ask
- a woman her age
- a man his salary
- a robot deployer how many robots they’ve deployed for non-research use cases
Afik Cohen retweeted
The mistake here was overestimating the capabilities of existing learning algorithms.
It should be self-evident that if we want to manufacture robots using processes that are not exact, or if aspects of their bodies change over time, then there is no way around continual adaptation. Tendon-driven hands introduce both of these challenges.
When your robot bodies are not exact, then just finding a performant policy on one body is not enough. Maintaining performance over time and successfully transferring the same policy to multiple bodies are both non-trivial tasks. This is something that Keen's Physical Atari setup clearly demonstrated.
Historically, the robotics industry has sidestepped this issue by making robots with tight tolerances and by maintaining those tolerances. It seems like Figure is headed in that direction as well.
Ultimately, the right solution is to fix the learning algorithms, organize the knowledge of the robot as self-verifiable subproblems, and have the robot maintain the correctness of this knowledge on its own through continual learning. Everything else is just a band-aid.
The biggest engineering mistake I made at Figure was building a tendon-based hand
Our first hand design in 2022 was a tendon hand for our F.01 robot. At a high level, the tendon approach sounds appealing, which is why I chose it: you get more space for packaging actuators since the forearm is larger, potentially higher degrees of freedom, and it's biologically inspired
We built and manufactured this entire hand and tested it in early 2023. If you saw it in person, it was truly an engineering work of art. It turned out to be one of the worst engineering decisions I've made, maybe the worst in four years. Tendons are a complete local maximum, and that only becomes clear in hindsight, after exploring every other possible hand design. Figure is unique in that we've now built and tested several hand architectures, which gave us a clear sense of where to head
Figure is working on our next generation hand, and it's so much better than anything you can do with tendons, it's not even funny
