Interview, Podcast
Fin Moorhouse - Longtermism, Space, & Entrepreneurship
- The scope for human achievement is expected to be extraordinarily large, potentially far exceeding current visions, provided humanity navigates a currently "unsustainably dangerous" moment without making irreversible errors.
- AI alignment is identified as the critical bottleneck and prerequisite for any long-term success; achieving this transition is viewed as necessary to secure a "defensive advantage" that allows humanity to remain safe indefinitely and unlock the capacity for new ideas.
- A sequential optimization strategy is predicted to be more effective than simultaneous optimization: first generating profit through useful products, then directing those resources toward maximizing good, which is expected to "pan out well" compared to trying to balance both from the start.
- The total funds dedicated to the movement are projected to grow significantly due to the risk-neutral nature of the philosophy and the absence of diminishing returns when aiming to "make the world go well," potentially leading to the emergence of EA billionaires among young, tech-savvy entrepreneurs.
- The movement is expected to face risks related to "dilution" of its original motivations as it grows, potentially resulting in "greenwashing" or a loss of the attitude of taking "weird ideas" seriously.
- There is a concern that "secular ethics" is a relatively new field (approx. 30 years old) and that humanity might fail to prevent large-scale risks simply because they cannot yet articulate or conceptualize them, similar to not knowing the local wildlife on an unknown trip.
- Future technological expansion into space is anticipated to be constrained by the universe's expansion, eventually making certain regions unreachable and rendering civilizations independent, though current "space governance" norms are considered an unlikely safeguard for the far future.
- Space is not expected to serve as a sufficient hedge against extinction because existential risks like unaligned AGI or dangerous pathogens are not independent across locations, and the absence of Mutually Assured Destruction in space could incentivize "first strike" strategies.
- Personal identity views may shift significantly if the "many worlds view" is true, potentially making it difficult to maintain intuitive concepts of continuous self due to the dissolving nature of the self into a "cloud of me's."
- The speaker anticipates building a startup in 2024, viewing the concept as highly compelling, though a non-profit approach would be considered if a profitable path is unavailable.
- Podcasts are predicted to be highly effective for idea communication due to higher information transfer rates in low-context environments, with a recommendation to view early failures (such as the first seven episodes) as a freeing normalization.
- Community building efforts carry a risk of negative marginal utility if they lead to adverse selection, attracting individuals less interested in the movement's core goals, though physical hubs for in-person interaction are expected to be powerful sources of motivation.
- The "long reflection" is framed as a directional ideal, suggesting that "taking time" is preferable to radical options, yet AI alignment is seen as a prerequisite for this reflection to be effective.
- Arguments regarding physical constraints on economic growth, such as "maximum GDP per atom," remain largely unresolved, and "common sense morality" may prove more effective than targeted interventions in areas like global health where capitalism has historically succeeded.
- A failure mode exists where current AI alignment research, potentially accelerated by initial grants, could inadvertently "speed up AI doom" if the transition is mishandled, creating an ambiguity where "common sense" might currently be safer.
- While reducing variance via global consensus is a goal, it carries the risk of cutting off expected value by excluding the "long tail" of possible best-case outcomes where the far future is exceptionally positive.
- The for-profit sector offers a tight, useful feedback loop where quick failure is possible, whereas the non-profit sector often lacks default mechanisms, creating a need for market-creating strategies like advanced market commitments to drive efficiency.
- Non-profit routes are expected to remain significant for areas lacking private market incentives, such as helping future generations or animals, due to "free rider" dynamics that prevent private investment.
- The speaker holds a strong prior that mistakes within the movement's theoretical or applied ideas can be identified within a reasonable timeframe given the movement's age, yet fears that the movement might underrate the potential for it to grow considerably and become mainstream.