AFFINE II Seminar Application — January 2027
Sign in to Google to save your progress. Learn more
Email *
AI-generated responses, or responses substantially inspired by AI, give us less information about how you reason and will carry less weight in our evaluation.
Name *
Email *
Seminar availability *
The seminar runs January 4–29, 2027. Can you attend the full seminar, arriving no later than January 3 and departing no earlier than January 30? Type "yes" if you can. If not, please explain your availability and circumstances. You are still welcome to apply if you cannot attend for the full period.
Choose one of the six prompts below to answer.
1. Justified assurance of friendliness
Suppose a mind X is significantly more intelligent than you by any reasonable definition of intelligence.

Is it possible to identify types of evidence that would justify confidence that X will continue to behave toward you in a friendly way indefinitely, even if you cannot understand how the strategies it adopts benefit you?
2. Empowering program
Do you expect there to exist a powerful (as in, has very large effects on the world) "empowering" program, such that (almost) any agent X would be better off, and does not regret, if they ran this program (as in let a computer execute the code).

If yes, give whatever detail you have about what such a program would look like. If not, please outline the reasons that make this difficult or impossible.
3. Triangulating between descriptions
What do you think about these two descriptions? What sorts of things do you think they are trying to point at? How do they differ? What do they have in common?

(1) Working backwards from future outcomes to possible choices that can produce those outcomes.

(2) Mediating causal influence in a way that enables some events to have large-scale effects that they otherwise would not have.
4. Neglected perspectives
Do you know any areas of research that we might be neglecting and that could contribute to our understanding of AI alignment or related phenomena? How would you incorporate them into alignment research?
5. Perspective on teleology
Teleology—understanding something in terms of the purpose it serves or is meant to serve—appears in many fields under various disguises. Which examples of teleological systems and discussions of teleological phenomena do you consider most relevant to the development of future artificial intelligences? What insights could these offer current AI alignment research about how to direct such intelligences toward appropriate purposes?
6. Philosophy for agency
What philosophical traditions have dealt with the ideas of (or related to) optimization and agency? How could these traditions be drawn on in order to help us better understand these topics?
Which question would you like to answer? *
Select one of the six prompts above, then answer it below.
Your response to the selected question *

Generally, aim to keep your response under 1,000 words. This is a guideline, not a hard limit: if you have lots of ideas, we want to hear them. However, we also value the ability to distill your thinking to its core insights, so avoid using more words than necessary.

Next
Clear form
Never submit passwords through Google Forms.
This form was created inside of cloutcareers.com.

Does this form look suspicious? Report

Google Forms