The Sequence Knowledge - Issue 941: Learning RSI: The Model Is Frozen. The System Is Not.
Your agent got better this quarter and nobody trained anything. Where did the improvement go? Into the notes. This is the loop that runs between model releases, and it might be where the durable value
An AI agent can improve over time even if its model and weights remain unchanged. This improvement stems from sources beyond the model itself, residing in other layers of the system. Understanding these different layers and their update economics is crucial for recognizing how AI capabilities advance.
- AI agents can improve noticeably over a quarter without any technical learning or model updates.
- Improvement in AI systems has three potential locations: the model (weights), and two other layers.
- The mechanics of these other two layers are often underrated and can lead to user-visible improvements.
- While the model’s weights receive significant attention, the top layer of the system receives most of the updates.
https://bender.layer3.press/articles/6217fde2-3cbb-4627-a25a-5d56faafffc9
Write a comment