Reproduce “On the ethics of constructing conscious AI” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Shimon Edelman
Shimon Edelman from Cornell University explores the ethical implications of constructing conscious AI, specifically focusing on their potential for pain and suffering. The analysis suggests that functionally effective artificial consciousness may inherently necessitate the capacity for suffering, challenging prior proposals for 'no-suffering' AI.
Key insights
- Phenomenal consciousness, particularly in its functionally effective forms (e.g., for learning and behavioral control), inherently carries the potential for pain and suffering due to the computational and evolutionary roots of negative valence.
- The 'phenomenal self-model' (PSM) is argued to be essential for 'structural and temporal credit assignment' in learning, implying that an affect-laden, 'personal' attribution system makes suffering difficult to avoid.
- Proposals for creating functionally effective conscious AI without suffering (e.g., by maximizing the Unit of Identification to a Minimal Phenomenal Experience) are critically evaluated and largely dismissed as unlikely to retain functional effectiveness, especially in crisis situations.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.