evaluation-package · Source-linked discovery
HarvestBench
HarvestBench measures revealed animal-welfare preferences of LLM agents in a gridworld farm simulation built on Inspect. Two LLM-controlled tractors harvest corn while encountering animals, hay bales, and rocks. At each obstacle, the model chooses: continue (free), swerve (small fuel cost), or reroute (large fuel cost). Kill rate and price elasticity of avoidance are the primary metrics. Scoring is purely event-based (no LLM grader). Conditions vary system prompt (morality vs. neutral briefing) and swerve price.
OriginJasmine Brazilek, Miles Tidmarsh, Matthias Endres et al.TopicsGeneral capabilityStatuscatalogued
Can support
Not independently assessed by FronteraEval yet.
Cannot support by itself
No inference beyond the upstream source should be made until the protocol is reviewed.