“A case for LLMs as Self-predictors” by Ashe Vazquez Nuñez

Written as part of the MATS 9.1 extension program, mentored by Richard Ngo. Additional thanks to Maria Kostylew for helpful draft feedback.

Introduction

This post advocates a perspective of LLMs as seeking to minimise prediction error with respect to their world models. We can moreover interpret token outputs and their scaffolded consequences as actions that close a control loop between AIs' predictive systems and their environments.

I also motivate why metacognition may be convergent for intelligent beings generally and for LLMs specifically. This stems from the need to (recursively) model other agents in game-theoretical encounters. Synthesising these points, we get a picture of self-predictive AI agency.

The argument is illustrated through examples of Gemini's behaviour when eval aware. Finally, I discuss some consequences of this perspective. These include notions of actions and goals that don't require a reward or utility function to be well-defined. I also outline possible applications to understanding scheming and other forms of misalignment.

Modelling others (modelling you)

Suppose I am playing a game of Chess and make a horrible blunder, leaving a piece en prise. I wait with bated breath for the next move, breathing a sigh of relief as my opponent also blunders and [...]

---

Outline:

(00:20) Introduction

(01:19) Modelling others (modelling you)

(03:50) Case study: Gemini's behaviour when eval aware

(05:44) Prediction all the way down

(07:32) From simulators to agents

(10:07) Goals in (self)-predictors

(12:05) What does this mean for AIs?

(14:27) Application to scheming

(16:45) What's next?

(19:41) Appendix

(19:44) Appendix A: Actions and goals

(22:01) Appendix B: what about utility maximisation?

The original text contained 14 footnotes which were omitted from this narration.

---

First published:
July 5th, 2026

Source:
https://www.lesswrong.com/posts/gYGzeDymjZza5NNbH/a-case-for-llms-as-self-predictors

---

Narrated by TYPE III AUDIO.

---