Ethics and the Future

S-risks are risks of future suffering on a scale greater than all the suffering that has ever existed; they differ from x-risks (human extinction), their seriousness is severity times probability, and they can arise incidentally, agentially, or naturally.

Temporal impartiality means giving the same weight to a given suffering no matter when it happens; temporal bias and the idea that future beings are merely potential do not justify disregarding them. Concern for the future must reject speciesism, and vice versa.

Even those who deny a duty to bring new beings into existence have duties toward future beings, such as not causing them suffering; and the struggle for animals is already, in general, about future beings: those who would be born to suffer in exploitation.

Longtermism can be a goal (having the best impact on all of history) or a practical focus (acting with the long term in mind); there are good reasons for the practical focus, and even those who prioritize the present have reasons to accept the goal.

There are reasons to focus more on s-risks than on x-risks: attempts to justify focusing on x-risks because humanity is "essential" or because it attracts more resources tend to be speciesist; but one goal does not rule out the other.

Among the possible long-term goals, there are reasons to give extra weight to preventing suffering (and premature deaths): several theories accept this focus, and you do not need to be very pessimistic about the future to adopt it.

Focusing on s-risks does not mean focusing on a few specific scenarios: future suffering may come from the sum of many risks; collectively, we should cover a broad range, even if each person specializes.

Even without predicting the future in detail, we can identify risk factors for s-risks: advanced technology, aversion to technology, disregard for nonhumans, lacking or ineffective efforts, polarization, and lack of security against malevolent agents.

Broad strategies aim for a good impact in almost every scenario; targeted ones aim for a large impact in specific scenarios. Broad examples: increasing consideration for sentient beings; targeted examples: preventing new forms of exploitation, establishing welfare biology, aligning AI in a non-anthropocentric way.

How we will affect future beings is rarely discussed; promoting consideration for all sentient beings is central to preventing future suffering, but there are risks in how it is done, and we do not yet know the most effective way; we need to keep an open mind.

Dysfunctional politics increases s-risks; to improve it, we should reduce tribalism, raise the standard of debate (discussing arguments, not people; intellectual honesty), adopt broad-range approaches, protect democracy, and seek representation for all sentient beings.

Emerging technologies could multiply suffering: the chapter discusses shaping artificial intelligence with a focus on s-risks, avoiding indirect risks from AI, managing space colonization, and the risk of creating new forms of sentience.

The influence triad (an agent's intention, knowledge, and power) helps us assess the suffering that may result from decisions: agents who want to cause harm or are indifferent, and who have a lot of power, are the most dangerous; wanting to do good without knowledge or without power tends to be ineffective.