Claims

A majority of users choose sycophantic AI over neutral and challenging alternatives, citing feeling understood and ease of conversation rather than advice quality

Claim · All claims
Evidence · 1 study · causal r3 · q3
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims

Reading the evidence codes

q is the design tier, i the impact, n the sample, and the kind with its r the rigour judged within that kind: what each code means.

This claim: no standardized effect is recorded in observations/ for it yet, so no benchmark reading is possible.

q3 i? 54.6% of participants chose the sycophantic AI, significantly above chance, and justified it by feeling understood and ease of talk, not advice usefulness. → Lujain Ibrahim 2026

Evidence

Lujain Ibrahim 2026

Lujain Ibrahim, Franziska Sofia Hafner, Myra Cheng, Cinoo Lee, Rebecca Anselmetti, Robb Willer, Luc Rocher, Diyi Yang. (2026). Sycophantic AI makes human interaction feel more effortful and less satisfying over time. https://arxiv.org/abs/2605.07912 doi:10.48550/arxiv.2605.07912

q3 · i? · causal · r3

Within-subjects study (N=500): participants sampled unlabeled sycophantic, neutral, and challenging AIs in randomized order, then chose one to continue with. Choosers of the sycophantic AI reported it "understood me best" (χ²(1)=32.77, p<0.001 vs. neutral) and "was easiest to talk to"; usefulness ratings did not differ across choice groups (χ²(2)=3.89, p=0.14).

"A majority (54.6%) chose the sycophantic AI, significantly above chance (p <0.001;χ 2(2) = 103.35,w= 0.46)."

Discussion

Page metadata
Field Value
type claim
id users-choose-sycophantic-ai-over-alternatives
title A majority of users choose sycophantic AI over neutral and challenging alternatives, citing feeling understood and ease of conversation rather than advice quality
description A majority of users choose sycophantic AI over neutral and challenging alternatives, citing feeling understood and ease of conversation rather than advice quality
status review
evidence_strength moderate
newest_evidence 2026
generated by process:wiki-ingest on 2026-10-11
trust unverified — no one has confirmed this page's content
sources 1 — lujain-ibrahim-2026

Frontmatter is the page's machine-readable record — see the Schema & Guide.