A collection of shorter pieces by Toby Ord.
(for all my academic papers, see Research)

Artificial Intelligence

How Well Does RL Scale?
I show that RL-training for LLMs scales much worse than inference or pre-training.

Evidence that Recent AI Gains are Mostly from Inference-Scaling
I show how most of the recent AI gains in reasoning come from spending much more compute every time the model is run.

The Extreme Inefficiency of RL for Frontier Models
The new RL scaling paradigm for AI reduces the amount of information a model could learn per hour of training by a factor of 1,000 to 1,000,000. What follows?

Is There a Half-Life for the Success Rates of AI Agents?
The declining success rates of AI agents on longer-duration tasks can be explained by a simple mathematical model — a constant rate of failing during each minute a human would take to do the task.

Inference Scaling Reshapes AI Governance
The shift towards inference scaling may mean the end of an era for AI governance. I explore the many consequences.

Inference Scaling and the Log-x Chart
The new trend to scaling up inference compute in AI has come hand-in-hand with an unusual new type of chart that can be highly misleading.

The Scaling Paradox
The scaling up of frontier AI models has been a huge success. But the scaling laws that inspired it actually show extremely poor returns to scale. What’s going on?

Existential Risk

The Precipice Revisited
How has the existential risk landscape changed in the 4 years since The Precipice came out?

On the Value of Advancing Progress
I show how a standard argument for advancing progress is extremely sensitive to how humanity’s story eventually ends.

Robust Longterm Comparisons
The choice of discount rate is crucially important when comparing options that could affect our entire future. Except when it isn’t.

The Timing of Labour Aimed at Reducing Existential Risk
Is it better to work on risks close to when they would occur, or as soon as possible?

Miscellany

Casting the Decisive Vote
An elegant proof that unless the election is a foregone conclusion, the chance your vote decides the whole thing can’t be much lower than one in the number of voters.

A Child’s Plaything
What a simple toy could tell the past about their future.

Remembering Peter Eckersley
A eulogy for Peter.