Research

{PAIR} Research Seminar

The research seminar runs weekly in the semester. The schedule is on the Research Seminar page.

Research Themes

At PAIR, we work on philosophical issues that relate to AI. Among these, we try to identify the most important and the most interesting ones. Our work is strongly influenced by empirical and engineering disciplines. At the moment, the areas for research that seem to have sufficient value to work on include the following. This list is undergoing continuous change.

Ethics of AI

  • Moral status of AI-systems, artificial moral agency & patiency
  • The AI alignment problem
  • Machine ethics (possibility, role of virtue ethics, rules, …)
  • AI ethical advisors, ethical enhancement, super-ethics
  • Superintelligence and XRisk (assumptions, prediction, policy, …)
  • Risks and Chances of AI in society (productivity, scientific & engineering progress, labor market, legal system, intellectual property rights, monopolies, fair distribution of gains, ecology …)
  • Impact of AI on values, justice, virtues
  • AI and responsibility allocation (responsibility gaps)
  • Decision-making with AI (human-in-the-loop, on-the-loop, bias, fairness, involvement of normativity …)
  • AI for deception and manipulation (including fake texts, imagery, video)
  • AI for surveillance (big data analysis, sensors, brain surveillance, “chilling” effects, …)
  • Risks from generative AI, e.g. large language models
  • Human dependence on AI, manipulation of humans, loss (or change) of “humanity”
  • Trust, trustworthiness
  • Can or should progress in AI be stopped or slowed down?
  • Ethical recommendations for regulations of AI
  • Political & societal structures to direct AI development and use

Theoretical Philosophy and AI

  • Models of natural cognition & neuroscience in relation to AI
  • Notion of computation
  • Notion of intelligence
  • Artificial sentience, role of consciousness in cognition
  • Role of normative cognition in intelligence
  • Normative approaches in AI evaluation, training, and reward (limits of rational choice, Markovian reward, etc.)
  • Normative approaches to AI interpretation problems (opacity, explainable AI)
  • Role of representation in AI (esp. concepts & language)
  • Use of AI in scientific research (e.g. neuroscience, medicine, …)
  • Epistemic issues generated by AI
  • The chances and limits of ML, relation of the to scaling of models
  • Embodied cognition and AI
  • Perception and the predictive mind
  • Philosophy through AI (esp. philosophy of mind, language, epistemology, conceptual engineering)

See also the research under people, and the activities on http://sophia.de/research.htm