# Leslie Kaelbling argues robotics is not solved and compositionality may help

> Original title: Leslie Kaelbling’s IROS 2026 workshop talk, “Small models for big worlds.” Leslie always gives contrarian views.

- Source: X search: AI launch posts (introducing, just launched)
- Published: 2026-11-09T18:25:08.000Z
- Bad Influence: https://badinfluence.news/items/a3xpu10byj17xldhs1l90naja
- Original: https://x.com/LinfengZhaoZLF/status/2119858844364898566

## Summary

Leslie Kaelbling's IROS 2026 workshop talk, "Small models for big worlds," argues robotics still has hard open problems that compositionality and modularity may help solve. She says Astra performs well at robotics for reasons including a compositional natural-language model of the world and recursive tool calls. The notes also list what an intelligent autonomous robot still needs, such as resource-rational computation and persistent memory.

## Post

Leslie Kaelbling’s IROS 2026 workshop talk, “Small models for big worlds.” Leslie always gives contrarian views.

Here’s a summary of her take on why compositionality helps, why Astra is so good at robotics, and why robotics is still not solved. https://x.com/i/article/2119749100769775617