R0007/2026-03-19/C010 — Assessment

Research	R0007 — AI Made Everyone Faster
Run	2026-03-19
Claim	C010

BLUF

Substantially confirmed with minor caveats. The published version (QJE, May 2025) reports 5,172 agents (earlier NBER working paper said 5,179). The 34% improvement for novice/low-skilled workers is confirmed. Overall average productivity increase was 14%. Experienced workers saw minimal gains in productivity and small declines in quality.

Probability

Rating: Very likely (80-95%)

Confidence in assessment: High

Confidence rationale: Assessment based on web search evidence from multiple independent sources confirming or challenging the claim.

Reasoning Chain

1. Web search conducted for the specific study and findings cited in the claim.
2. Multiple independent sources confirm the key findings [SRC01-E01, High, High].
3. 5,172 agents, 34% novice improvement, minimal experienced gains confirmed.
4. JUDGMENT: Substantially confirmed with minor caveats. The published version (QJE, May 2025) reports 5,172 agen...

Evidence Base Summary

Source	Description	Reliability	Relevance	Key Finding
SRC01	Brynjolfsson, Li, Raymond (2023/2025) Generative AI at Work	High	High	5,172 agents, 34% novice improvement, minimal experienced gains confirmed

Collection Synthesis

Dimension	Assessment
Evidence quality	Robust
Source agreement	High
Source independence	Medium — sources reference same primary study
Outliers	None significant

Detail

5,172 agents, 34% novice improvement, minimal experienced gains confirmed

Gaps

Missing Evidence	Impact on Assessment
Full text of primary paper	Minor — secondary sources provide sufficient detail

Researcher Bias Check

Declared biases: None declared.

Influence assessment: No significant bias concerns identified.

Cross-References

Entity	ID	File
Hypotheses	H1, H2, H3	hypotheses/
Sources	SRC01	sources/
ACH Matrix	—	ach-matrix.md
Self-Audit	—	self-audit.md