Steve Abreu

Dhaval Adjodah Verified account @_dval_ · Nov 3

Running AI is expensive! @AnthropicAI's amended S-1 today shows $895B committed to compute and infrastructure but with Q3 revenue $17B and $567M loss, mostly from short-term compute.
@makermaker_ai can make it cheaper, better and faster with RSI. makermaker.ai/breakthroughs/… 🧵1/7

View quoted post on X

a16z @a16z · Oct 12

MakerMaker.AI co-founders Dhaval Adjodah and Owen He with a16z's Martin Casado on recursive self-improvement, and how to tell when it's real:

A lot of RSI pitches are really just people using AI to build better AI. Tech has always worked that way. You needed computers to build chips. Steam engines got better once you could get coal faster.

Dhaval and Owen's bar is higher. The model has to improve itself: its weights, its harness, even the prompt that tells it what to work on next.

And the gains have to hold up when you count what they cost. They track everyone's experiment spend on a "bleeder board," and one engineer, the Viceroy of Variance, has the job of breaking their results.

Over the past few months, they've started seeing early signs of their system getting better and cheaper at the same time.

0:55 Why most RSI pitches miss
2:25 Optimizer vs. optimizee
5:10 The RSI equation: performance over cost
6:50 Why a closed loop does not mean takeoff
8:40 Humans are agents inside the system too
9:25 The denominator trap: progress divided by capital raised
11:40 Training GPT-2 for $100
13:40 Most RSI papers fail when replicated
16:40 Why you have to co-optimize
17:40 Why inference is the first problem
20:15 The agent rewrites its own prompt
22:20 "It's not a belief. It's happening."
25:00 Why they haven't published
27:50 Intelligence is a vector and Amdahl's law
29:45 The garden hose theory of RSI

YouTube: youtu.be/3Q1mdn5ZGvM

@makermaker_ai @_dval_ @martin_casado

View quoted post on X