{"type":"post","id":"kaggle-and-rehearsals-launched-human-behavior-bench-scoring-0fd223","slug":"kaggle-and-rehearsals-launched-human-behavior-bench-scoring-0fd223","title":"Kaggle and Rehearsals launched Human Behavior Bench, scoring LLMs on predicting real human survey distributions.","description":"Human Behavior Bench makes a specific claim testable: a model can sound plausible about people yet fail to reproduce how a defined group distributes its answers.","retrieval_nugget":"Human Behavior Bench makes a specific claim testable: a model can sound plausible about people yet fail to reproduce how a defined group distributes its answers.","published_at":"2026-10-21","updated_at":"2026-10-21","record_date":"2026-10-21","date_kind":"published_at","topics":["evals","models"],"source_urls":["https://www.runrehearsals.com/human-behavior-bench"],"schema_version":"newruntime-agent-readable-v0.2","stable_id":"post:kaggle-and-rehearsals-launched-human-behavior-bench-scoring-0fd223","status":"published","visuals":[],"editorial_provenance":{"schema_version":"newruntime-editorial-copy-v1","content_status":"source_grounded_final","final_copy_sha256":"sha256:406a32649de0d50fabdac59f1c15e5e9e56d185d715fbb09f583212555174b87","reviewed_at":"2026-10-21T08:05:08.924Z","source_evidence_count":1,"verified_claim_count":3,"site_analysis_schema_version":"newruntime-site-analysis-v1","site_object_kind":"field_note","observed_fact_count":3,"implication_count":2,"watch_condition_count":1,"related_record_count":0},"analysis":{"schema_version":"newruntime-site-analysis-v1","object_kind":"field_note","thesis":"Audience-level prediction needs evaluation against measured distributions, not a fluent account of what people might do.","observed_facts":[{"text":"The benchmark reports 1,199 tasks based on surveys of 1,244 U.S. adults, with 239 public and 960 private tasks.","source_urls":["https://www.runrehearsals.com/human-behavior-bench"]},{"text":"The published table lists six LLM scores between 76.2 and 80.6, and Rehearsals reports 94.1 for Engram 1.0.","source_urls":["https://www.runrehearsals.com/human-behavior-bench"]},{"text":"The source says base models ran through Kaggle's harness while Engram used Rehearsals' in-house evaluation system, with the same case and metric setup.","source_urls":["https://www.runrehearsals.com/human-behavior-bench"]}],"mechanism":"The mechanism is to compare each model's predicted answer distribution with survey ground truth for a defined cohort, rather than grade a free-form explanation.","why_now":"The benchmark makes a gap between conversational plausibility and audience-level prediction legible as a scored evaluation, including in product and marketing questions.","implications":["The first implication is that teams should not treat a model's confident audience persona or preference summary as equivalent to measured user research.","Evaluation results may identify domains, such as product choices, where model-generated audience estimates need stronger validation."],"evidence_boundary":"The evidence boundary is that These are organizer-reported results; base models used Kaggle's harness and Engram used Rehearsals' in-house harness, albeit with the same case and metric setup. The page does not establish independent replication, representativeness beyond the U.S. panel, or predictive validity for every business decision.","watch_conditions":["The first watch condition is whether independent evaluators reproduce the rankings and whether performance generalizes across new populations, survey modes, and decision domains."],"related_records":[],"new_branch_reason":"No prior New Runtime record is supplied for this specific benchmark result; it establishes a distinct analytical record about measuring audience-level prediction against survey distributions."},"editor":{"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]},"routes":{"html":"https://newruntime.com/posts/kaggle-and-rehearsals-launched-human-behavior-bench-scoring-0fd223/","markdown":"https://newruntime.com/posts/kaggle-and-rehearsals-launched-human-behavior-bench-scoring-0fd223.md","json":"https://newruntime.com/posts/kaggle-and-rehearsals-launched-human-behavior-bench-scoring-0fd223.json"},"related_patterns":["ai-engineering-evaluations-are-splitting-into-task-specific"],"next_reads":[{"type":"topic","path":"/topics/evals/","url":"https://newruntime.com/topics/evals/","title":"Evals","reason":"Explore the evals topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/evals.md","json_url":"https://newruntime.com/topics/evals.json"},{"type":"topic","path":"/topics/models/","url":"https://newruntime.com/topics/models/","title":"Models","reason":"Explore the models topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/models.md","json_url":"https://newruntime.com/topics/models.json"},{"type":"collection","path":"/posts/","url":"https://newruntime.com/posts/","title":"Field Notes","reason":"Open the latest editorial analysis.","media_type":"text/html","json_url":"https://newruntime.com/posts.json"},{"type":"collection","path":"/source-ledger.json","url":"https://newruntime.com/source-ledger.json","title":"Source Ledger","reason":"Inspect the public source evidence graph.","media_type":"application/json"}]}