Publications (AI / society)

This is a selected list of publications focusses on sociotechnical AI research and AI safety. You can find a full listing of our work on Google Scholar.

2025 / In Press

• • 
  • How will advanced AI systems impact democracy?
  • Summefield et al
  • Nature Human Behaviour (in press)
• • 
  • Deep reinforcement learning can promote sustainable human behaviour in a common-pool resource problem
  • Koster et al
  • Nature Communications (2025)
• • 
  • Deep mechanism design: Learning social and economic policies for human benefit
  • Tachetti et al
  • PNAS (2025)
• • 
  • Reward Model Interpretability via Optimal and Pessimal Tokens
  • Christian et al
  • FacCT Proceedings (2025)
• • 
  • Why human–AI relationships need socioaffective alignment
  • Kirk et al
  • Humanities and Social Sciences Communications (2025)
• • 
  • Conversational AI increases political knowledge as effectively as self-directed internet search
  • Luettgau et al
  • arXiv
• • 
  • Technological folie a deux: Feedback Loops Between AI Chatbots and Mental Illness
  • Nour et al
  • arXiv
• • 
  • The Levers of Political Persuasion with Conversational AI
  • Hackenburg et al
  • arXiv
• • 
  • Lessons from a Chimp: AI 'Scheming' and the Quest for Ape Language
  • Summerfield et al
  • arXiv
• • 
  • How Malicious AI Swarms Can Threaten Democracy
  • Schroeder et al
  • arXiv
• • 
  • Increasing happiness through conversations with artificial intelligence
  • Heffner et al
  • arXiv (2025)
• • 
  • Language agents as digital representatives in collective decision-making
  • Jarrett et al
  • arXiv (2025)
• • 
  • Can AI Mediation Improve Democratic Deliberation?
  • Tessler et al
  • Knight First Amendment Institute (2025)

2024

• • 
  • AI can help humans find common ground in democratic deliberation
  • Tessler et al
  • Science (2024)

2023

• • 
  • Using the Veil of Ignorance to align AI systems with principles of justice
  • Weidinger et al
  • PNAS (2023)

2022

• • 
  • Fine-tuning language models to find agreement among humans with diverse preferences
  • Bakker et al
  • NeurIPS (2022)
• • 
  • Human-centred mechanism design with Democratic AI
  • Koster et al
  • Nature Human Behaviour (2022)
• • 
  • The good shepherd: An oracle agent for mechanism design
  • Balaguer et al
  • arXiv
• • 
  • Hcmd-zero: Learning value aligned mechanisms from data
  • Balaguer et al
  • arXiv