Expert opinions on Future

Even after 3 years working on AI capability evaluations, the recent pace of progress has been a shock. To avoid catastrophic harms from AGI, we need much, much better coordination.

Much has been said about whether AI poses an ultimate existential threat to humanity. The more immediate existential challenge may be whether technologies that simulate relationships weaken our capacity for the relationships that foster our own human...

By framing AI as a mysterious entity, an existential threat, a utopian savior, or a moral scapegoat, these narratives obscure the reality of AI as a statistical tool.

This creates what we call the “Abundance Flywheel”: commit compute to a hard problem, focus R&D until it clears, watch the domain collapse from artisanal craft to industrial utility, capture the surplus, and reinvest it into the next target. Repeat u...

AI could make goods and services cheaper than ever, raising hopes for a new age of abundance where scarcity plays a smaller role in human life.

It is often suggested that the advent of powerful AI will create abundance and that it can thereby contribute to securing social harmony. This paper draws on Hirsch’s study of different kinds of limits to growth to show that this is false.

MAP-AI, a new control-plane system architecture for aligned decision-making under uncertainty, enforces alignment through Monte Carlo estimation of outcome distributions and admissibility-controlled policy selection rather than static model-level con...

If it takes over that's what will happen. [...] We don't know how to do that. We don't even know if it's possible.

No amount of computational power or rationality can avoid intrinsic alignment overheads.

The UK must lead an international agreement to prohibit the development of superintelligence until it can be proven safe. If we fail, every other ambition for our country risks being swept aside. The clock is ticking.

The question that the noble Lord posed was: do we therefore go for a moratorium? That would be highly desirable, but I do not think that it will be possible in the short term.

What is needed is not to slow innovation but to harness it — to channel AI’s momentum toward outcomes that reinforce trust, safety and societal well-being.

The greatest of those risks is not intelligence or what the technology can do, but the concentration of power in too few companies, in too few countries and, ultimately, in too few hands.

Assuming capabilities continue to accelerate, one responsible course may be for frontier developers to stop, investigate, and let alignment research and governance catch up.

Because the point of AI safety is not to slow the future down. It is to make sure the future remains human, aligned with our values, and advancing our interests.