AI Safety

MichaelDickens maps 3 AI futures for sentient welfare work strategies

Three possible ASI timelines dictate how to prioritize sentient welfare efforts

Deep Dive

MichaelDickens outlines three possible futures based on AGI/ASI timelines and alignment success, each with distinct implications for sentient welfare work. In a long timelines scenario (decades+), there is room for foundational research in moral philosophy, decision theory, moral circle expansion, theoretical AI alignment paradigms, and traditional animal advocacy. In a short timelines scenario where alignment is on track (e.g., powerful aligned ASI soon), the priority shifts to steering that ASI to care about sentient welfare—through research on aligning values, making LLMs more animal-friendly, and targeting frontier AI developers with animal welfare arguments. In the most pessimistic short-timelines scenario (alignment unsolved), all other work is moot; the only priority is preventing catastrophic misalignment via AI pause advocacy, safety regulations, and alignment research.

The post emphasizes that many strategies implicitly bet on one future, and individuals should probe their assumptions. At the community level, however, diversification across futures is wise: some should work on long-term foundational problems, others on optimistic short-term alignment steering, and others on safety advocacy. The piece also raises the question of which plausible futures might be neglected entirely—a topic for further exploration. Overall, it provides a framework for resource allocation in the AI safety and sentient welfare space.

Key Points
  • Long ASI timelines enable foundational work like moral philosophy, decision theory, and traditional animal advocacy.
  • Short timelines with alignment success shift focus to steering ASI toward sentient welfare (e.g., aligning LLMs, influencing AI developers).
  • Short timelines without alignment demand AI pause advocacy and safety regulation as the single priority.
  • Community should diversify across futures rather than betting entirely on one scenario.

Why It Matters

Offers a structured lens for AI safety professionals to allocate effort across different timeline beliefs.