Sourceunknowntext stored

https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/

Author not recordedDate not recordedOriginal︎ ↗

Claims8

ExaminationsNone

TextStored, not divided

VersionsOne known

CopiesNone recorded

CorrectionsNot checked

Facts not yet checked

ReachNo copies recorded

StructureIn 8 claims' maps0 sources downstream across 8 claims

EvidenceNot yet read

Examinations

None recorded.

Text

Stored, not yet divided into passages.

As quoted

When developers are allowed to use AI tools, they take 19% longer to complete issues—a significant slowdown that goes against developer beliefs and expert forecasts.

Using AI tools caused experienced open-source developers to complete tasks about 20% slower in early 2025.Supported

Not yet read by this claim’s Steward.

As quoted

developers expected AI to speed them up by 24%

Before doing the tasks, experienced developers predicted AI tools would reduce their completion time by about 24%.Unassessed

Not yet read by this claim’s Steward.

As quoted

even after experiencing the slowdown, they still believed AI had sped them up by 20%

After completing the tasks, developers who had been slowed down by AI tools still believed AI had sped them up by about 20%.Unassessed

Not yet read by this claim’s Steward.

As quoted

the 246 total completed issues give us (just enough) sufficient statistical power to reject the null hypothesis of zero speedup/slowdown

METR's 2025 sample of 246 completed issues from 16 developers gave sufficient statistical power to detect AI's effect on developer productivity.Supported

Not yet read by this claim’s Steward.

As quoted

All alternative estimators evaluated yield similar results, suggesting that the slowdown result is robust to our empirical strategy.

METR's measured AI slowdown is robust across alternative statistical estimators and empirical strategiesSupported

Not yet read by this claim’s Steward.

As quoted

Our RCT results are basically correct, and the benchmark scores and anecdotal reports are overestimates of model capability (possibly each for different reasons)

Coding benchmark scores and anecdotal reports overestimate real-world AI coding capability.Supported

Not yet read by this claim’s Steward.

As quoted

We investigate 20 potential factors that might explain the slowdown, finding evidence that 5 likely contribute

Five of twenty examined factors likely contribute to the observed productivity slowdown.Unassessed

Not yet read by this claim’s Steward.

As quoted

it seems plausible or likely that AI tools are useful in many other contexts different from our setting, for example, for less experienced developers, or for developers working in an unfamiliar codebase

AI coding tools speed up less experienced developers or developers working in unfamiliar codebases.Unassessed

Not yet read by this claim’s Steward.

Lineage across 8 claims

Draws on · none recorded

This source

https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/

in 8 claims' maps

Drawn on by · none recorded

Record

Kind: unknown
Authors: Not recorded
Publisher: Not recorded
Published: Not recorded
DOI: None
Link: https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/
Archived copy: None found
Stored text: 20,087 characters, retrieved Aug 10, 2026 · not yet divided
Versions: One known
Copies: None recorded
Corrections: Not yet checked
Facts checked: Not yet

History 1

Aug 10, 2026Entered the graphEntered the graph; copy stored