FLF’s Focus

What we do

Enormous resources are being expended in racing toward powerful autonomous AI systems. Some, concerned about hazards such as superintelligence, aim to slow, redirect, or prevent this race. With more powerful AI coming – in some form and at some rate – we’re trying to make humanity more capable of steering AI’s direction wisely and managing its effect.

We look for high-impact gaps in the landscape — important work that isn’t getting the resourcing and attention it needs. Then we make it happen: starting organizations, running fellowships, recruiting founders, funding projects. We don’t accept funding applications. We research what matters and go get it built.

Our primary focus: AI tools for human collective agency

Technology shapes the world we live in today. The technologies we develop now — especially AI-powered technologies — will shape the world of tomorrow.

We are concerned that humanity may fumble this ball. High stakes and rapid, dynamic, changes mean that leaders and other decision-makers may be disoriented, misunderstand the situation, or fail to coordinate on necessary actions — and steer the world into gradual or acute catastrophe.

The right technology could help. The rise of modern AI systems unlocks prospects for big improvements here, in ways that were out of reach just a couple of years ago, and almost unthinkable five years ago. AI tools might help everyone track the state of play, make decisions they stand behind, and act in sync with others. This could defuse race dynamics, prevent narrowly-interested cliques from exploiting less coordinated groups, and avoid catastrophic errors of judgement.

Thus far, it seems very few people are building the systems that could unlock collective wisdom and coordination. This may simply be inertia — people haven’t woken up to the possibilities. It is also market incentives at play: AI is unlocking a lot, and these aren’t necessarily the applications which stand to make the most money in the short term.

We are therefore especially excited to see tools in areas like:

Collective epistemics — determining what’s true, evaluating evidence, tracking the provenance of arguments and data

Coordination — overcoming collective action problems, building trust and commitment mechanisms, surfacing preferences and combining them into shared goals, finding agreements closer to the utility frontier

Planning & forecasting — anticipating consequences, modeling complex systems, giving decision-makers more time to think before they have to act

Underlying these is a general principle: to look for things that will augment and improve rather than replace human judgment. Stronger AI may generate incentives (or even imperatives) to fully automate human judgment in some areas. Sometimes this could be acceptable, but we think it is unlikely to have good outcomes unless people were able to make informed decisions about whether and when to do so — informed decisions which can be substantially enabled in the meantime by good AI tools for collective agency. We’re not sure humanity is on track to enable this important choice, so we aim to help build these tools to do so.

Although our primary focus is on these technologies, this is because we think there may be especially high-leverage opportunities there today. Our broader mission remains making humanity more capable of steering AI. When we see high-impact, timely, neglected opportunities in areas like governance, institutional capacity-building, or safety infrastructure, we’ll pursue those too.

Notes on urgency

AI capabilities are advancing fast. If capabilities feed back into AI development, this could further accelerate things. If humanity handles that poorly, this could lead to the end of everything we have valued.

While we cannot be certain about the future, it is our view that it is important to plan for a sudden explosion of capabilities, quite possibly in the 2020s. This doesn’t leave us much time! We want to move quickly, and to support projects that are well-positioned to accelerate as the world accelerates.

For more on our background views, see our post: AI Timelines and How They Inform FLF’s Plans and Work