Executive summary: The author argues that animal welfare concerns should be dominated by post–artificial superintelligence (ASI) futures in which humans survive, since even well-aligned outcomes under a coherent extrapolated volition (CEV) framework could still allow large amounts of animal suffering depending on how human values are extrapolated and implemented.
Key points:
The author posits a future where an ASI aligned to humanity’s CEV governs outcomes, and explores how this could affect animal welfare.
They note that CEV relies on extrapolated human preferences, which may vary widely and depend on arbitrary features of the extrapolation procedure.
They highlight that some humans’ extrapolated volitions might include preserving natural ecosystems with live, unmodified animals, which could perpetuate animal suffering.
The author reviews Eliezer Yudkowsky’s framing of CEV and Bostrom’s parliamentary model, including Thomas Cederborg’s critique that the “random-dictator” baseline empowers harmful or “troll” agents.
They argue that even principled alignment approaches may still yield futures with animal suffering, given the difficulty of principled trade-offs and coherence in value aggregation.
The best prospects for animal welfare improvement, they suggest, lie in avoiding unaligned ASI and refining philosophical and meta-philosophical understanding to better specify extrapolation procedures.
The author concludes that while these outcomes are unsatisfying, animal welfare’s fate likely hinges on how successfully alignment and value extrapolation are handled in post-ASI futures.
This comment was auto-generated by the EA Forum Team. Feel free to point out issues with this summary by replying to the comment, andcontact us if you have feedback.
Executive summary: The author argues that animal welfare concerns should be dominated by post–artificial superintelligence (ASI) futures in which humans survive, since even well-aligned outcomes under a coherent extrapolated volition (CEV) framework could still allow large amounts of animal suffering depending on how human values are extrapolated and implemented.
Key points:
The author posits a future where an ASI aligned to humanity’s CEV governs outcomes, and explores how this could affect animal welfare.
They note that CEV relies on extrapolated human preferences, which may vary widely and depend on arbitrary features of the extrapolation procedure.
They highlight that some humans’ extrapolated volitions might include preserving natural ecosystems with live, unmodified animals, which could perpetuate animal suffering.
The author reviews Eliezer Yudkowsky’s framing of CEV and Bostrom’s parliamentary model, including Thomas Cederborg’s critique that the “random-dictator” baseline empowers harmful or “troll” agents.
They argue that even principled alignment approaches may still yield futures with animal suffering, given the difficulty of principled trade-offs and coherence in value aggregation.
The best prospects for animal welfare improvement, they suggest, lie in avoiding unaligned ASI and refining philosophical and meta-philosophical understanding to better specify extrapolation procedures.
The author concludes that while these outcomes are unsatisfying, animal welfare’s fate likely hinges on how successfully alignment and value extrapolation are handled in post-ASI futures.
This comment was auto-generated by the EA Forum Team. Feel free to point out issues with this summary by replying to the comment, and contact us if you have feedback.