# METR

METR studies what AI systems can do and how reliably they complete tasks, including work with human developers.

Type: organization. Context selected by Escape Meridian; inclusion is not endorsement.

[Collection](/entities/metr)
[Original website](https://metr.org/)

## Sources

- [METR](https://metr.org/): Compare benchmark capability with task completion, reliability and actual practitioner workflows.

## Perspectives


## Terminal notes

- [Coding assistance changes faster than its measurement](/notes/developer-productivity-measurement.md): METR reports that selection effects and concurrent agent use make its newer developer experiment unreliable for estimating current speedup. It considers improved productivity plausible while declining to assign a dependable effect size.

## Parallels

- [More capable. More in control?](/parallels?parallel=delegated-judgment): When does handing over the work give people more room to decide?
- [The answer is right. Was the question?](/parallels?parallel=right-answer-wrong-question): When producing an answer gets cheaper, where does the useful work move?

## Context references

- [https://metr.org/](https://metr.org/)

Latest published distillations: [Publications API](/api/pipeline/publications?entity=metr)