Neural steering vectors reveal dose and exposure-dependent impacts of human-AI relationships

Documenti analoghi

• Why human-AI relationships need socioaffective alignment
  di: Kirk, Hannah Rose, et al.
  Pubblicazione: (2025)
• PRISM-X: Experiments on Personalised Fine-Tuning with Human and Simulated Users
  di: Kirk, Hannah Rose, et al.
  Pubblicazione: (2026)
• Conversational AI increases political knowledge as effectively as self-directed internet search
  di: Luettgau, Lennart, et al.
  Pubblicazione: (2025)
• SimpleSafetyTests: a Test Suite for Identifying Critical Safety Risks in Large Language Models
  di: Vidgen, Bertie, et al.
  Pubblicazione: (2023)
• When Do LLM Preferences Predict Downstream Behavior?
  di: Slama, Katarina, et al.
  Pubblicazione: (2026)