Andrej Karpathy says the gap in understanding what LLMs can do is widening into a sharp funnel
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.
Context
One of the most respected voices in AI argues that only a tiny sliver of people see what frontier models can really do, which helps explain why public debate about AI capability feels so disconnected.
Karpathy is a founding member of OpenAI and the former head of AI at Tesla. In April he posted that there was a growing gap in understanding of AI capability, partly because many people judged AI from free tier chatbots. This new post quotes that April thread and argues the gap has since grown.
Deep dive
On October 11, Andrej Karpathy revisited an April post about a gap in how people understand AI capability, saying it is now less two groups speaking past each other and more a sharp funnel.
By his napkin math, about 6 billion people have barely come in contact with LLMs at all, and 1 to 2 billion are casual users of free tier chatbots who treat them like a better search engine or writing aid. Around 20 million people use frontier grade models professionally in math and code and see agents complete projects that used to take weeks or months. At the narrow end, he estimates about 5,000 people with internal access to frontier systems, where he says swarms of agents collaborate for weeks on large software projects, find zero day exploits and advance mathematics, while human review starts to fall behind.
Karpathy says three factors drive the funnel: impact scales with the ambition and size of the problem, capability is jagged and peaks in digital, verifiable and economically valuable domains because of how models are trained with reinforcement learning, and access differs across free, paid and internal tiers. He concludes that a small sliver of professionals are experiencing the vertigo of the curve going vertical while most people still see a derpy chatbot.
What to expect next
Watch whether labs move to narrow the funnel by bringing more capable models to free tiers, and whether the gap between internal and public systems becomes a policy issue as calls for oversight of frontier labs grow.
Also watch for data that tests Karpathy's estimates, such as usage figures from OpenAI, Anthropic and Google on how many people use agents for long, complex work.