Pat Grady and Sonya Huang at Sequoia Capital:
An AI that can figure things out has some baseline knowledge (pre-training), the ability to reason over that knowledge (inference-time compute), and the ability to iterate its way to the answer (long-horizon agents).
The first ingredient (knowledge / pre-training) is what fueled the original ChatGPT moment in 2022. The second (reasoning / inference-time compute) came with the release of o1 in late 2024. The third (iteration / long-horizon agents) came in the last few weeks with Claude Code and other coding agents crossing a capability threshold.
Generally intelligent people can work autonomously for hours at a time, making and fixing their mistakes and figuring out what to do next without being told. Generally intelligent agents can do the same thing. This is new.
More here.
Enjoying the content on 3QD? Help keep us going by donating now.