{
 "@context": "https://schema.org",
 "@type": "Claim",
 "@id": "https://wulfkaal.github.io/claims/4855607-015",
 "identifier": "kaal:claim:4855607-015",
 "text": "Reward modeling learned through interaction with users carries two structural pathologies: majority views disproportionately influence the learned reward function, and the agent may engage in reward hacking.",
 "author": {
  "@type": "Person",
  "name": "Wulf A. Kaal",
  "identifier": "https://orcid.org/0000-0003-0757-275X"
 },
 "citation": "Wulf A. Kaal, How AI Models are Optimized Through Web3 Governance (2024). SSRN: https://ssrn.com/abstract=4855607",
 "isBasedOn": {
  "@type": "ScholarlyArticle",
  "name": "How AI Models are Optimized Through Web3 Governance",
  "datePublished": "2024",
  "url": "https://ssrn.com/abstract=4855607",
  "sha256": "eb0b3e62374b45a8fa888c6bde9725e606bcb46cf4b5e74a6e851d9f25099113",
  "encoding": "https://raw.githubusercontent.com/wulfkaal/Academic-Papers/main/papers/pdf/Kaal%20-%202024%20-%20How%20AI%20Models%20are%20Optimized%20Through%20Web3%20Governance.pdf"
 },
 "keywords": [
  "ai-and-agents"
 ],
 "about": [
  "reward-modeling",
  "majority-bias",
  "reward-hacking",
  "generative-ai",
  "preference-aggregation"
 ],
 "abstract": "but it comes with potential issues such as the prevalence of majority views disproportionately influencing the learned reward function and the risk of reward hacking.",
 "additionalProperty": [
  {
   "@type": "PropertyValue",
   "name": "claim_type",
   "value": "failure"
  },
  {
   "@type": "PropertyValue",
   "name": "confidence",
   "value": "evidenced"
  },
  {
   "@type": "PropertyValue",
   "name": "scope_conditions",
   "value": [
    "reward functions learned from aggregated user interaction"
   ]
  },
  {
   "@type": "PropertyValue",
   "name": "is_failure_mode",
   "value": true
  },
  {
   "@type": "PropertyValue",
   "name": "attestations",
   "value": "https://wulfkaal.github.io/colloquium/attestations/61a867481c493ba4a1f5ef08fd9f012f1e8e03104a49e07af84e625734ea8a34.json"
  },
  {
   "@type": "PropertyValue",
   "name": "edges",
   "value": []
  }
 ],
 "isPartOf": {
  "@id": "https://wulfkaal.github.io/claims/index.json"
 },
 "dateModified": "2026-07-27",
 "version": "1.0",
 "sha256": "61a867481c493ba4a1f5ef08fd9f012f1e8e03104a49e07af84e625734ea8a34",
 "canonicalForm": "https://wulfkaal.github.io/claims/4855607-015.md"
}