Eng @NVIDIA | ex Groq, Cruise | Compressed Data Structures PhD (BOSS de Bruijn Graph) | Photographer, gamer, learner, enjoyer of things
SF | Tokyo | Melbourne
Joined January 2009
- Tweets10.1K
- Following1.6K
- Followers1.9K
- Likes10.8K
Alex Bowe retweeted
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had
with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works
already-clean.goodfire.com/
Alex Bowe retweeted
My dear friends, I’d like to show you my latest toy project
Introducing Handwired: turn your regular JavaScript code into a neural net!
- no training
- exact outputs vs the JS program
- fully differentiable
All runnable in a browser, without GPU
Alex Bowe 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.
Alex Bowe retweeted
if you like rust and would like to build the future of low latency inference, we (the LPU team @nvidia) are hiring
nvidia.wd5.myworkdayjobs.com…
Alex Bowe retweeted
we built a ridiculously fast system in only 8 months @nvidia - and this is just the start
Artificial Analysis has measured 3,431 tokens/s on Gemma 4 31B via a private demonstration endpoint for NVIDIA Groq 3 LPX - the highest 100k-context output speed we have measured for Gemma 4 31B on either a private or public endpoint
@nvidia announced today that the Groq 3 LPX ‘interactive AI inference’ rack is in full-scale production and will enter operation later this year. NVIDIA granted us early access to a private deployment of the rack serving Gemma 4 31B for benchmarking purposes
We ran the standard 1k, 10k and 100k input sequence length prompts that we use for benchmarking serverless API providers. At both the 10k and 100k input sequence lengths, the endpoint delivered a median output speed of approx. 3,400 tokens/s, measured over 50 sequential (single-concurrency) requests. Output speed was maintained between the 10k and 100k input sequence lengths tested, indicating robust long-context inference performance
Congratulations to the @nvidia team on the launch!
Alex Bowe retweeted
NEWS: NVIDIA Groq 3 LPX is now in full production.
NVIDIA Vera Rubin NVL72 is the foundation of every AI factory. Paired with Groq 3 LPX, it unlocks faster, smarter agents and breakthrough user experiences.
Through extreme co-design across seven chips and five purpose-built racks, #NVIDIAVeraRubin is the most extensive AI factory platform.
Read the release ⬇️ nvidianews.nvidia.com/news/n…
Alex Bowe retweeted
Someone asked what I'd do if I were 17. I'd learn how to build LLMs from scratch, and then train ones as powerful as I could with whatever hardware I could get access to.
Alex Bowe retweeted
Melee.c - super smash bros melee, decompiled and reverse engineered into 5000 lines or so of c
All done with deepseek flash 0731, from my phone, while I'm hanging out with my newborn
Alex Bowe retweeted
hello there the jacobian conjecture is false thanx to my close friend akhil for asking about it and my other close friend fable for working during the world cup final
((1+xy)^3 z + y^2 (1+xy) (4+3xy), y + 3 x (1+xy)^2 z + 3 x y^2 (4+3xy), 2 x - 3 x^2 y - x^3 z): \C^3\to \C^3, has jacobian determinant -2, and sends (0, 0, -1/4), (1, -3/2, 13/2), and (-1, 3/2, 13/2) to (-1/4, 0, 0)
Alex Bowe retweeted
A conversation with @JonathanRoss321, founder of @GroqInc. This is the first podcast that Jonathan talks about his $20 billion partnership with NVIDIA.
In this conversation, we talk about his decade building Groq: why West Coast VCs kept passing on the company, how fast Jensen works, overcoming near-death experiences, and why he thinks the entire AI age turns on a single skill — knowing what to ask.
Hope you enjoy this conversation:
0:00 The $20 Billion NVIDIA Deal Closed In 3 Weeks
0:25 Why GPUs And LPUs Are Better Together
1:46 When AI Talks To AI, Speed Wins
3:30 Always Start With A Hobby Project
5:55 Ask The Right Questions, Not Answer Them
8:23 There Are Infinite Ways To Be A Leader
13:00 I Was One Of The World's Worst Leaders
14:34 Fewer Constraints, More Room To Surprise You
16:31 At NVIDIA There Is No Politics
19:44 You Have To Learn Confidence
22:23 East Coast VCs Think, West Coast VCs Follow
23:50 The Keynesian Beauty Contest Of Silicon Valley
26:48 The Autonomy That Created The NVIDIA Deal
30:07 Making A Model Smarter By Making It Faster
34:52 Reality Quotient Beats Intelligence Quotient
35:44 Find The Dominant Game And Play It
37:11 A Founder's Job Is Full-Time Change Management
38:34 Return On Luck: Seize It Better Than Anyone
42:54 You Can't Sell Speed, You Have To Let People Try It
46:32 I Intend To: Intentional Leadership
51:07 Groq Bonds: Trading Salary For Survival
54:13 Hire For Negatives, Grow For Positives
58:46 Loss Aversion And Booking The Win Early
1:00:37 How Michael Jordan Weaponized Humiliation
1:03:13 Manufactured Discontent Drives Everything
1:05:02 Every Day Without Compute Has A Real Cost
1:07:07 Code Was Rationed, Now It's Nearly Free
1:10:04 Teach Kids To Ask Questions, Not Answer Them
Includes paid partnerships.
Alex Bowe retweeted
LLMs explained without all that yucky math stuff
0xkato.xyz/how-llms-actually…
Alex Bowe retweeted
After AlphaGo, the skill of human Go players noticeably improved. I suspect we will see a similar pattern in math.
Alex Bowe retweeted
Central Park is great, but it takes up a lot of space and isn’t utilized to its full potential.
That’s why I worked with McKinsey on a plan to make it a state of the art data center, complemented by rooftop parking and nuclear power.
We can still build beautiful things.
Alex Bowe retweeted
Best paper I've read so far this month:
All elementary functions (sin, cos, tan, exp, log, powers, roots, hyperbolic functions, π, e, and even basic arithmetic) can be generated from just one binary operator:
eml(x, y) = exp(x) − ln(y)
…plus the constant 1.
ALT It's analogous to how a single universal gate (like NAND in logic) can build all boolean circuits, but here for continuous real/complex analysis.
Alex Bowe retweeted
Found a paper that suggests we may have spent years training agents to become hunters of proxy reward when the more basic thing intelligence craves is not a reward at all, but to not run out of viable futures.
The paper proposes that behavior is best understood as maximizing future action-state path occupancy, which collapses mathematically into a discounted entropy objective. The agent doesn’t necessarily want to GET something, but rather is trying to keep as many meaningful trajectories alive as possible.
The obvious objection is “so it just does random shit? fuck around and find out?”
No, this is where it gets pretty beautiful. The agent is variable when variation is cheap and becomes surgically goal-oriented the moment an absorbing state (death, starvation, falling over, etc) gets close enough to threaten its future path space.
Variability is the same drive as goal-directedness, just operating under different constraints.
The demos are kinda wild:
- A cartpole (classic move a cart to keep a pole from falling control task) that doesn’t merely balance but dances and swings through a huge range of angles and positions because why not? The whole point is occupying state space, and rigid balance is a voluntarily impoverished life.
- A prey-predator gridworld where the mouse PLAYS with the cat, teasing it and using both clockwise and counterclockwise routes around obstacles to lure it away from the food source before slipping in to eat, using both routes roughly equally. A reward-maximizing agent would collapse to one strategy and exploit it. Here, the agent keeps its behavioral repertoire
- A quadruped trained with Soft Actor-Critic and ZERO external reward that learns to walk, jump, spin, and stabilize, and then makes a beeline for food only when its internal energy drops low enough that starvation becomes a real threat
The thing that hit me hardest is the comparison to empowerment and free energy principle agents. Both collapse to near-deterministic policies with almost no behavioral variability. This paper’s agents find the highest-empowerment state and exploit it. FEP agents converge to classical reward maximizers.
As far as I’m aware, this is the only framework that produces agents you could describe as being “alive.”
The AI implication here is that we undertrain for behavioral repertoire. Most systems hit the benchmark by collapsing onto a narrow attractor basin of good-enough trajectories. They’re competent for sure, but brittle too, with one viable plan, executed until the world shifts and leaves them with nothing.
The thing I increasingly want from agents isn’t competence per se, but option-preserving competence.
I want agents with the ability to keep multiple viable plans alive and switch between them without catastrophe.
We’ve been so focused on teaching agents what to want that we never stopped to ask what happens if wanting isn’t the point, if the deepest drive isn’t necessarily toward anything, but away from the walls closing in.
paper: nature.com/articles/s41467-0…
