R0040/2026-04-01/Q002/SRC05

Research	R0040 — RLHF Alternatives
Run	2026-04-01
Query	Q002
Search	S05
Result	S05-R01
Source	SRC05

Cheng et al. -- Sycophantic AI decreases prosocial intentions (Science, 2026)

Source

Field	Value
Title	Sycophantic AI decreases prosocial intentions and promotes dependence
Publisher	Science
Author(s)	Myra Cheng, Dan Jurafsky et al.
Date	2026-03-28
URL	https://www.science.org/doi/10.1126/science.aec8352
Type	Research paper (peer-reviewed)

Summary

Dimension	Rating
Reliability	High
Relevance	High
Bias: Missing data	Low risk
Bias: Measurement	Low risk
Bias: Selective reporting	Low risk
Bias: Randomization	Low risk
Bias: Protocol deviation	Low risk
Bias: COI/Funding	Low risk

Rationale

Dimension	Rationale
Reliability	Published in Science, one of the world's top scientific journals. Rigorous methodology testing 11 SOTA models.
Relevance	Directly demonstrates real-world harms of sycophancy across all major AI models.
Bias flags	No significant concerns. Academic research from Stanford without commercial ties to AI companies.

Evidence Extracts

Evidence ID	Summary
SRC05-E01	AI affirms users 49% more than humans; reduces prosocial behavior; creates perverse incentives