“AI’s capability improvements haven’t come from it getting less affordable” by Anders Woodruff

Mar 27, 2026

A data-driven look at whether AI progress is becoming less affordable, using METR time-horizon trends and a cost-ratio definition. Clear breakdowns of how frontier models perform at 50% reliability and whether longer tasks drive gains. Discussion of fixed-cost horizons, inference-scaling effects, methodological limits, and why this analysis differs from other cost estimates.

Ask episode

AI Snips

Chapters

Transcript

Episode notes

INSIGHT

Rising Inference Costs Reflect Longer Tasks

Anders Woodruff argues rising inference costs reflect models tackling longer tasks, not becoming more expensive relative to human labor.
Meta's frontier models hit 50% reliability horizons at ~3% of human cost, a stable ratio across model generations.

INSIGHT

Cost Ratios At 50% Horizon Aren't Increasing

Woodruff measures cost ratio as per-task AI inference cost divided by human cost and finds no upward trend at each model's 50% time horizon.
Using Meta's time-horizon data, median cost ratios at those horizons remain well below 1 across successive models.

INSIGHT

Successful Long Tasks Aren't Pricier Per Hour

Among tasks models successfully complete, longer tasks are not systematically more expensive per hour than shorter ones.
Selection effects exist (models pass fewer long tasks), but successful long tasks tend to be relatively cheap.

Get the Snipd Podcast app to discover more snips from this episode

Get the app

METR's frontier time horizons are doubling every few months, providing substantial evidence that AI will soon be able to automate many tasks or even jobs. But per-task inference costs have also risen sharply, and automation requires AI labor to be affordable, not just possible.[1] Many people look at the rising compute bills behind frontier models and conclude that automation will soon become unaffordable.

I think this misreads the data. The rise in inference cost reflects models completing longer tasks, not models becoming more expensive relative to the human labor they replace. Current frontier models complete tasks at their 50% reliability horizon for roughly 3% of human cost, and this hasn’t increased as capabilities have improved.

I define cost ratio as the inference cost of the average AI trajectory that solves a task divided by human cost to complete the same task. Using METR's data, I examine the trend in cost ratios over time. I show three things:

Across successive frontier models, the cost ratio at each model's 50% reliability time horizon hasn't increased.
Among tasks models successfully complete, longer tasks don't have higher cost ratios than shorter ones.
Time horizon trends barely slow when capping AI spending per [...]

---

Outline:

(02:42) Evidence from METR

(03:22) Cost ratio at models 50% time horizon isnt increasing

(04:59) Time horizons improvements arent driven by expensive long tasks

(06:07) Progress at a fixed cost is just as fast

(08:04) Limitations of my methodology

(09:42) Inference scaling will just make progress faster

(11:10) Conclusion

(12:12) Appendix A: Why I get different results from Ord

(15:32) Appendix B: Additional cost at time horizon graphs

(16:45) Appendix C: 80% affordable time horizon