Interview
Peter Singer: Suffering in Humans, Animals, and AI | Lex Fridman Podcast #107
- Anticipates that while objective suffering might be eliminated, relative forms such as boredom and lack of purpose are likely to persist in affluent societies, with a "little drop of poison" remaining in the human experience even in a harmonious state.
- Predicts that current AI lacks a "subject of experiences" despite mimicking consciousness, yet foresees machine learning creating autonomous systems where the emergence of fear or consciousness becomes undetectable across billions of nodes, necessitating a "benefit of the doubt" approach for beings with ambiguous consciousness.
- Notes that failing to grant rights to conscious robots could lead to a hardened human disposition toward other suffering beings like animals, while acknowledging a theoretical trade-off where a minimum of suffering might be required to experience positive highs.
- Identifies significant risks in AGI development, including the potential for accidental loss of control and existential extinction risks which Singer estimates are vast (e.g., $10^{56}$) rather than infinite, with the view that humanity will be in a "much better position" to address these threats 80 years after a 100-year timeline.
- States that democratic systems are expected to remain the preferred method for ethical decision-making due to the unlikelihood of a benevolent leader avoiding corruption, while global convergence on a single utilitarian "right answer" is deemed difficult due to differing intuitions about justice.
- Highlights that calculating optimal utilitarian outcomes is hindered by the inability to measure happiness quantities or action probabilities, though he expects effective altruism to provide greater life fulfillment and satisfaction than traditional career paths by connecting individuals to larger purposes.
- Observes that ethical standards regarding animal treatment and rights are expected to shift over generations, with societal acceptance of previously considered ridiculous concepts and a historical pattern where what one generation accepts, the next may shudder at.
- Estimates that very few people could truly claim they would not have participated in atrocities like the Holocaust if placed in similar circumstances, with fewer than 10% of the population likely able to assert this.
- Acknowledges the theoretical possibility of constructing consciousness that experiences pleasure while automatically cutting out suffering via "instant anesthesia," though confirms such technology does not currently exist on Earth.
- Admits that while he hopes humanity can "scrape through" climate change with less damage, he does not expect to live long enough to witness the full outcomes of future environmental conditions.