FAQ

At the camp you can:

• Meet insightful and diversely competent individuals dedicated to AI Safety.
• Learn about research concepts, practices and mindsets in the field.
• Deliberate how to research an open problem with a mentor.
• Find more interested collaborators to work with over the months and years ahead.
• Be more systematic and productive in developing research as part of a team.
• Learn what it’s like to do research in a team.
• Test your personal fit to inform your next career steps.
  
  But how does AISC help me develop my career?

  Rephrasing points above, AI Safety Camp offers you the opportunity to

  • get in touch with others who will update you on interesting developments and new opportunities in the community.
  • learn concepts and jargon that allows you to communicate better with future recruiters.
  • make yourself useful to a mentor who wants to support people making progress on their problem.
  • try working with new and potentially long-term research collaborators.
  • learn beyond the basics by actually researching a problem that interests you.
  • find out how much you enjoy and are able to contribute to this research on the frontiers.
  • publish work that helps established research organizations better discern where you may be able to contribute.
  • get some 1-on-1 coaching to consider next career steps (incl. for roles in operations, entrepreneurship, or earning-to-give/save).

We expect you will:

• Communicate your ideas and uncertainties clearly and succinctly, and ensure your arguments are well grounded in existing research literature.
• Be proactive in reaching out to others at AISC, whether it be mentors, your team, other participants or AISC organizers. Ask for help when you’re stuck or wonder if you’re going off course.
• Do your background research before reaching out to mentors. Brainstorm, edit, and clarify your thoughts either individually or as a team. Consider what the mentor is particularly well-positioned to give feedback on and direction for. In general, be respectful of mentors’ time.
• Regularly step back to evaluate and assess what could make your project even more of a success.
• Consider that the falsification of hypotheses is also a research success.
• Be invested in your team’s growth and results. You’re not alone in this, you’re a team!

People with disabilities are invited to apply to AISC and can be assured that reasonable accommodations will be made. After admission, Sai Joseph, our participant facilitation coordinator, will invite you to submit documentation of your disability as well as any suggestions you have for how we can assist you to get the best out of the camp. Sai will then look into you situation and what accommodations we can offer.

We appreciate all efforts that contribute to safer AI.
→ To collaborate on addressing present-day problems in society, we recommend you check out other research communities such as Fairness, Accountability, and Transparency in ML or Robust and Responsible AI.

→ To try collaborating on research to safeguard society’s long-term trajectory for the common good of all, apply to AISC.
AI developments pose existential risks to future generations. If an existential risk comes to pass, society’s trajectory will end. You may have reason to prioritise other systemic risks though.

To learn about our long-term perspective on safety research, check out the 80,000 Hours Podcast or AI Alignment Podcast. Each interview comes with transcripts and links to mentioned papers, if you prefer to read!

Step 1: Apply

Draft in this doc. We value nuanced honest answers. Do also make a case for yourself.
Copy over and send in the writing by the deadline – 1 December 11:59 pm any time on earth – or earlier to hear back faster. Late submissions might not get a response.

Step 2: Wait for 1-2 weeks

Four of our reviewers each independently rate one section of your application (drawing from their own involvement at AISC and in AI Safety research). They then come together to discuss aggregate ratings and remaining uncertainties around how to interpret your writing. If we conclude that you might pass our criteria, one of us will email an invitation for an interview (or else, a gentle declination). We aim to get back to you within two weeks.

Step 3: Book an interview

One of us will invite you to a 45-minute video call. Please be prepared and on time.
You will have only five minutes to ask questions about the camp, so try to reach out to us beforehand.

While our final evaluation mostly hinges on your written answers, the interview gives you an opportunity to clarify and collaboratively discuss your ideas real-time. We will get back to you with an admit/decline decision within the ensuing week.

Look through the answers you gave on the application form (especially on your selected research problem) and make sure you can answer follow-up questions we might ask to clarify your reasoning further. Rehearse a couple of times.

Prepare 5 minutes before the interview by pointing a light source at your face, putting the camera at eye level, and checking that your microphone is working – to ensure we can follow what you’re saying.

Five weeks. We intend to send you an invitation to the camp (or a declination) within five weeks after the day you submit your application (note: we increased this period from three to four and then to five weeks given a larger-than-expected number of applications).

Attendance is free. If the camp experience helped you make next career steps, do signal this value to us with a ‘pay what you want’ donation afterwards. This pays for organisers to run next editions.

AI Safety Camp is geared towards diversely skilled people who want to test their fit by collaborating intensively on research in their spare time. Internships (and to a lesser extent, fellowships) at established AI Safety organizations select for people who have advanced further in their studies, can convey their considerations and assumptions using concepts and jargon of the core community, and can work on research full-time.

You will join research weekend sprints online rather than our usual 9-day mountain retreat. We have designed the 2022 virtual program so you can have insightful discussions and work flexibly within your timezone. See the schedule here.

From January to June 2022 expect to commit >1 h/d on average during normal workdays and >7 h/d during the scheduled weekend sprints Meaning that over the six months you will usually spent ~5 hours over a week on project work, and once/twice a month you will spent ~19 hours over the week (note: after final presentations on 4 June, you’ll still need to wrap up your project and write a doc explaining what your team did).

To join the camp, you’re expected to free up the above-mentioned hours in your schedule. What matters though is doing quality research together. If you can effectively collaborate to produce higher-quality research in less time, so the better! Of course, a few of the Mon-Fridays might not count as a regular workday for you and you can’t do work then. And if you have to, say, visit an older family member during a weekend sprint, just let us and teammates know and find some way to work around it.

Narrow down the scope of your research and try to not take up any further commitments than you already have. It’s common for graduating students to doubly underestimate the hours they’ll spent in the end – to complete their university thesis AND their research at the camp. This gets sucky for their AISC teammates when they become slow to respond or finish obvious tasks. Avoid being just another person who plans forward optimistically, by accounting for usual ways that projects like this will get side-railed for a while. To reliably estimate the time you’ll need to be able to spend on another commitment, recall any similar project in the past and tally up the hours that roughly took to complete.

Between team formation in mid-February and presentations in early June, teams will work together to finalize their research plan, complete any background reading, perform their research, and work to present their results to the rest of the camp and potentially to the public.

Teams will plan weekly meetings with either their mentor or with the mentoring coordinator, Adam Shimi. Additionally, we’ll have three co-working sprints, one on the first weekend of every month, to give a chance for each team to get together for dedicated focused work. Throughout the process, Adam will check in with the teams to help ensure that the project is going smoothly and the team is on track to be able to reach a satisfying completion.

At the start of June, every team will write a summary of what they have attempted and accomplished and then give a presentation. Of course, not every project is successful: many promising ideas don’t pan out, or require significantly more effort than originally hoped. Many teams continue their work after the camp, either to finish up any loose ends or to ready their work for publication in a journal or blog post.

These are established AI existential safety researchers who are happy to offer brief but regular guidance for researching their open problem. They are usually available for a weekly call, and to give feedback on your final draft. To make the most of your mentor’s time, prepare a short update for them on your progress and 1-3 specific questions you want to clarify ahead of each call.

These weekends are an opportunity for participants to focus on AISC work and make substantial progress on their projects. We expect participants to minimize other commitments and be able to devote 14-16 hours of work over the course of the weekend. These weekends are also tied to milestones in the research process in order to help ensure that teams are not falling behind the pace they need to reach a satisfying conclusion to their projects.

We start Saturdays together by organisers’ sharing announcements, senior researchers answering your questions, and/or teams presenting updates on their progress. On Sundays, we focus on discussing and co-working on research we’re interested in in smaller groups.

Once teams are formed, we will have a coworking sprint every first weekend of March, April and May.
You can mostly fill in your weekend how you like. However, you can expect a mixture of:

• Presenting an update on your team’s progress and bottlenecks, and asking other teams for input.
• Checking in briefly with Adam and planning your team’s next steps
• Coworking in our virtual office with timed pomodoro breaks
• Pair programming or other paired work where one creates and the other refines
• Joining spontaneous social activities

Yes, in most cases. Main sessions usually start at about 4pm UTC and sessions can continue until about 9pm UTC. This means people in continental Asia as well as New Zealand should be able to join fine if they can sleep a little later or get up a little earlier. But for Singapore, Indonesia and Australia much of those sessions fall in the middle of the night. We want to make sure we can cater to those timezones well but haven’t found a great solution yet (perhaps recording sessions + checking in later in a smaller group). Tips appreciated!

The short answer is yes. We want you to meet with new collaborators and motivate each other to work efficiently with a mentor.
Some listed problems require an unusual interest or specialist skills though (e.g. tabletop game design or evolutionary genetics). If you are the one person who applies and gets admitted to work on such a problem, then you are a team of one.

No, you won’t be able to participate in two teams at the same time. However, you’re very welcome to give feedback to another team or discuss their topic during breaks. Chats with other teams can inspire you to take a novel perspective on your own research.