It’s Tokens All The Way Down: How RLMs are Different — Kevin Madura, AlixPartners

Kevin Madura from AlixPartners presents recursive language models (RLM) — a method to make AI models work more like programmers than language processors.

Instead of attempting to read through all tokens directly, an RLM can treat its content as variables in a code environment, write Python code to solve problems, and even delegate to itself or other models to break down difficult tasks. On a benchmark for long reasoning, accuracy improved from 2.6 percent to 45.4 percent, particularly for tasks that become straightforward when written as code. He demonstrates practical examples such as analyzing customer data, summarizing long invoices, and finding patterns in logs — all without needing to break up the text into pieces.

More from AI Engineer

Inside Arca's Long-Horizon Agents That Run the Business - Aman Singh, Arca

AI Engineer · 23 Oct

Real AI Agent Attacks and What Would Have Stopped Them — Andrey Rybka, Bloomberg

AI Engineer · 23 Oct

Outside of tech there has been no financial payoff to US companies for using AI, now viewed as a margin-hurting cost center, as VC-backed firms burn more tokens

Techmeme · 4 h ago

Source: Replit recently moved most AI spend from Anthropic to OpenAI, which gave 30%+ off in a deal worth $100M+/year, after Anthropic discontinued its discount

Techmeme · 6 h ago