RESEARCH · RESEARCH · #2639
Meta-analysis: LLMs show modest loss aversion, mixed diminishing sensitivity, human-like probability weighting
The paper (arXiv:2610.33103v1) meta-analyzes 102 fitted estimates from 11 studies covering 32 language models to test prospect-theory components in LLMs. It finds modest loss aversion (pooled λ = 1.26, 95% CI 1.06–1.46), value functions that are near-linear or method-dependent for diminishing sensitivity, and human-like inverse-S probability weighting (γ ≈ 0.81–0.84) though based on fewer studies.
KEY POINTS
- The paper (arXiv:2610.33103v1) meta-analyzes 102 fitted estimates from 11 studies covering 32 language models to test prospect-theory components in LLMs.
- It finds modest loss aversion (pooled λ = 1.26, 95% CI 1.06–1.46), value functions that are near-linear or method-dependent for diminishing sensitivity, and human-like inverse-S probability weighting (γ ≈ 0.81–0.84) though based on fewer studies.
- This matters because it shows LLMs do not fully replicate human risk preferences—affecting their use and evaluation in decision-support, alignment, and safety-sensitive tasks.
WHY IT MATTERS
This matters because it shows LLMs do not fully replicate human risk preferences—affecting their use and evaluation in decision-support, alignment, and safety-sensitive tasks.