Rubi Hudson

Conditioning Predictive Models: Large language models as predictors

This is the first of seven posts in the Conditioning Predictive Models Sequence based on the paper “Conditioning Predictive Models: Risks and Strategies” by Evan Hubinger, Adam Jermyn, Johannes Treutlein, Rubi Hudson, and Kate Woolverton. Each post in the sequence corresponds to a different section of the paper. Thanks to Paul Christiano, Kyle McDonell, Laria Reynolds, Collin Burns, Rohin Shah, Ethan Perez, Nicholas Schiefer, Sam Marks, William Saunders, Evan R. Murphy, Paul Colognese, Tamera Lanham, Arun Jose, Ramana Kumar, Thomas Woodside, Abram Demski, Jared Kaplan, Beth Barnes, Danny Hernandez, Amanda Askell, Robert Krzyzanowski, and Andrei Alexandru for useful conversations, comments, and feedback. Abstract Our intention is to provide a definitive reference on what it would take to safely make use of generative/predictive models in the absence of a solution to the Eliciting Latent Knowledge problem. Furthermore, we believe that large language models can be understood as such predictive models of the world, and that such a conceptualization raises significant opportunities for their safe yet powerful use via carefully conditioning them to predict desirable outputs. Unfortunately, such approaches also raise a variety of potentially fatal safety problems, particularly surrounding situations where predictive models predict the output of other AI systems, potentially unbeknownst to us. There are numerous potential solutions to such problems, however, primarily via carefully conditioning models to predict the things we want—e.g. humans—rather than the things we don’t—e.g. malign AIs. Furthermore, due to the simplicity of the prediction objective, we believe that predictive models present the easiest inner alignment problem that we are aware of. As a result, we think that conditioning approaches for predictive models represent the safest known way of eliciting human-level and slightly superhuman capabilities from large language models and other similar future m

89Feb 2, 2023

Rubi Hudson

Message

823

205

Defining Monitorable and Useful Goals

In my most recent post, I introduced a corrigibility transformation that could take an arbitrary goal over external environments and define a corrigible goal with no hit to performance. That post focused on corrigibility and deception in training, which are some of the biggest problems in AI alignment, but the...

Jul 15, 202515

Defining Corrigible and Useful Goals

This post contains similar content to a forthcoming paper, in a framing more directly addressed to readers already interested in and informed about alignment. I include some less formal thoughts, and cut some technical details. That paper, A Corrigibility Transformation: Specifying Goals That Robustly Allow For Goal Modification, will be...

Jun 25, 202538

Safe Predictive Agents with Joint Scoring Rules

Thanks to Evan Hubinger for funding this project and for introducing me to predictive models, Johannes Treutlein for many fruitful discussions on related topics, and Dan Valentine for providing valuable feedback on my code implementation. In September 2023, I received four months of funding through Manifund to extend my initial...

Oct 9, 202455

Simplifying Corrigibility – Subagent Corrigibility Is Not Anti-Natural

Max Harms recently published an interesting series of posts on corrigibility, which argue that corrigibility should be the sole objective we try to give to a potentially superintelligent AI. A large installment in the series is dedicated to cataloging the properties that make up such a goal, with open questions...

Jul 16, 202445

The Case for Predictive Models

Thanks to Johannes Treutlein and Paul Colognese for feedback on this post. Just over a year ago, the Conditioning Predictive Models paper was released. It laid out an argument and a plan for using powerful predictive models to reduce existential risk from AI, and outlined some foreseeable challenges to doing...

Apr 3, 202443

Searching for Searching for Search

Thanks to Leo Gao, Nicholas Dupuis, Paul Colognese, Janus, and Andrei Alexandru for their thoughts. This post was mostly written in 2022, and pulled out of my drafts after recent conversations on the topic. My main update from then is the framing. Rather that suggesting searching for substeps of search...

Feb 14, 202421

Conditional Prediction with Zero-Sum Training Solves Self-Fulfilling Prophecies

Thanks to Charlotte Siegmann, Caspar Oesterheld, Spencer Becker-Kahn, and Evan Hubinger for providing feedback on this post. The issue of self-fulfilling prophecies, also known as performative prediction, arises when the act of making a prediction can affect its own outcome. Systems aiming for accurate predictions are then incentivized not only...

May 26, 202388

Load More (7/20)

AI ALIGNMENT FORUM
AF

AI ALIGNMENT FORUM
AF

Rubi Hudson

Rubi Hudson

Rubi Hudson

Conditioning Predictive Models: Large language models as predictors

Conditional Prediction with Zero-Sum Training Solves Self-Fulfilling Prophecies

Proper scoring rules don’t guarantee predicting fixed points

Conditioning Predictive Models: Outer alignment via careful conditioning

Rubi Hudson

Defining Monitorable and Useful Goals

Defining Corrigible and Useful Goals

Safe Predictive Agents with Joint Scoring Rules

Simplifying Corrigibility – Subagent Corrigibility Is Not Anti-Natural

The Case for Predictive Models

Searching for Searching for Search

Conditional Prediction with Zero-Sum Training Solves Self-Fulfilling Prophecies

Conditioning Predictive Models: Large language models as predictors

Conditional Prediction with Zero-Sum Training Solves Self-Fulfilling Prophecies

Proper scoring rules don’t guarantee predicting fixed points

Conditioning Predictive Models: Outer alignment via careful conditioning

Defining Monitorable and Useful Goals

Defining Corrigible and Useful Goals

Safe Predictive Agents with Joint Scoring Rules

Simplifying Corrigibility – Subagent Corrigibility Is Not Anti-Natural

The Case for Predictive Models

Searching for Searching for Search

Conditional Prediction with Zero-Sum Training Solves Self-Fulfilling Prophecies