Kareem 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.
swarm alignment != individual alignment
NVIDIA's agent safety stack is the right kind of fix: infrastructure, enforced outside the agent.
But testing every bee doesn't tell you what the hive does, and the same goes for agents, so we'll need new techniques for the swarm itself.
None of this requires anthropomorphizing the agents or assuming they want anything. Goals can drift at the swarm level through ordinary learned habits, and I explain how in this essay:
kareem.me/research/the-ai-ag…
p(doom | AI) ≥
p(has open-ended goals | AI)
× p(makes own subgoals | has open-ended goals, …)
× p(seeks power | makes own subgoals, …)
× p(misaligned | seeks power, …)
× p(can outmaneuver us | misaligned, …)
× p(oversight fails | can outmaneuver us, …)
× p(doom | oversight fails, …)
≥ means this is one path only, and misuse, gradual multipolar disempowerment, and AI-triggered conflict are separate chains that add on top.
Thought about the subgoals problem while listening to Geoff Hinton: youtu.be/N1TEjTeQeg0?t=1382
Professor Geoffrey Hinton discusses the concept of subgoals in the context of artificial intelligence threats starting at (23:05).
He explains that to create an intelligent agent capable of achieving complex tasks, it becomes necessary to allow it to generate its own subgoals (23:06-23:16). He expresses concern that such systems will quickly identify gaining more control as a universal subgoal, as it facilitates the achievement of almost any other objective (23:33-23:44)
What if recursive self-improvement doesn’t need backprop? Thousands of agents experimenting, sharing what works, and improving their own tools: an evolutionary-style loop around frozen models. The weights don’t change. But the system does. Oh… whoops.
Probably worst time to share my blog post about actual fly brain mechanics that I wrote back in July:
kareem.me/research/flys-tric…
Kareem retweeted
I didn’t think autonomy would get nationwide-scale this fast... but here we are 📷!
Our work with @HUMAIN and @TareqAmin_ will bring thousands of trucks to KSA. But this is just the start – followed by robotaxis, ports, mining, construction, and more.
The size, scale and scope is unprecedented! Letssss go! 🚀🚀🚀
Kareem retweeted
What's a meta-agent?
It's an agent that manages other agents mid-task. It can intervene, rewind, fork & audit them as they work – think "git for agents".
It's also the emerging layer between you & your agents, a new interface for human-AI interaction!
Here's what it unlocks🧵
TLDR: OpenAI is building a laptop where codex primitives are brought down to the OS.
1/OpenAI shipped a $230 macropad last week and many have filed it as a gimmick.
Look closer. imo the Codex Micro is the most legible public artifact of how OpenAI is thinking about the OS for their consumer devices.
🧵from someone who built OS + ~1B param models 👇
Kareem retweeted
Guys, guys, we only need 100k CoT rollouts to fix the LLM maff
xitter.cf/kimocode/status/208050…
Kareem retweeted
The knowledge that makes AI useful is diffused. It lives with scientists, engineers, clinicians, firms. For AI to benefit from distributed knowledge, it must itself be distributed. Agree with Jensen that this is a future worth building.
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.
AI will transform every industry, power every company, and be built by every country.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.
The world needs both frontier closed models and frontier open models.
images.nvidia.com/pdf/Open-W…
