Leonard Bereska Not you?

Iliad Intensive

October 2026

The most important things

Before
Join the Iliad Discord and check you can see #intensive-october. That is where you post any material you need to share with participants. If you can't see it, tell Leonard (@leonardbereska on Discord).
During
At 17:10, make sure everyone fills out the feedback form.
After
Fill out the end-of-day log: rate the day and tell us anything that went wrong.
Fill out Who stood out: name anyone who made a positive impression.
  • Questions about pay, travel, logistics or materials: see the FAQ.

66 project proposals are open for review. Stars, a sentence, or both. Read them.

Your days

Project proposal day · Wed 21 Oct Wed 21 Oct, leading 21 responses · energy 4.2/5
Overall +0.68 better than expected n = 19
Content +0.42 better than expected n = 19
Clarity +0.26 better than expected n = 19
Motivation +0.58 better than expected n = 19
Exercises +0.50 better than expected n = 8
Readings 0.00 as expected n = 4
Other activities +1.00 better than expected n = 12
Pace +0.21 about right n = 19
Difficulty +0.21 about right n = 19
Novelty +0.26 mostly new n = 19
Read what they wrote22
Worked well6
  • Today, it was the proposal day. It was very intense, I enjoyed it and it was very nice both the format of the activity and the synergy with other participants.
  • good sinergy with group mates
  • I found working on actual research proposal and running a small experiment very productive.
  • I think it was good having us write proposals and engage deeply with a particular topic.
  • It was great to have some time to dive deeper into a topic based on what we felt inspired by.
  • The sporadic collaboration on any research topic was surprisingly very fruitful! I had a lot of fun engaging with everyone and just seeing what problems people were interested in.
Would improve13
  • I think it’s amazing that we have a whole day to write a project proposal! I think the structure of the day works as well — first giving us time to write up the proposal, then elicit feedback/ peer review, and then finally submit it. This is not quite a critique of the format, but if we could have a little more time, perhaps until dinner or midnight to turn in the proposal as opposed to 5:30 it would be helpful! That way if we’re hoping to submit more than 1 proposal we have enough time to spend working with two different teams and turning it in! I think the time provided currently is definitely sufficient to write up one good proposal, I had to rush my second one but there was still enough time to get across the basic idea. It has been so fun discussing, writing, and reviewing ideas, definitely one of my favorite days!
  • I would have posted to guidelines the night before.
  • At better scoring system. The average of top two submissions means that if you have one good proposal and another you are not sure about then you don't necessarily want to submit the second one.
  • Maybe having a set of interesting papers that are recommended to work on would be good!
  • more time, id do something like 2 days.
  • The word counts of the proposal could have been clarified a bit more, especially as compression of ideas is often the hardest step. The point about how multiple projects would be scored was not clear as well.
  • In google docs the word count seems like a recommendation, while on the form it is a hard limit. Would have been good to know. Also to know the structure earlier in the intensive would have been better.
  • Announce the grading of project proposals well in advance and tell us to think about project during the intensive!
  • More feedback from experienced researchers. Clearer explanation of the intent. Was this an exercise or an exam?
  • I think 2 proposals was too much. It took me a long time just to get the 1 proposal right.
  • I felt that the peer review process could have been better organized by having someone come up and say it was time to review others' papers. Some were still brainstorming proposals or still writing when this happened which made it hard for those who had finished on time to incorporate review comments when the second wave actually got to them. Some also openly admitted to not reviewing anyone's proposals which felt a bit unaltruistic.
  • i have a bit of mixed feelings about this... on the one hand it's great to have time to think about a possible original contribution to the field, and I am glad to have gone through the exercise. on the other hand, it's a bit risky, from the academic-credit-attribution perspective, to assemble people into groups and have them pool ideas and pitch projects, without knowing whether people will have the opportunity to work together (say one person in a group gets into a fellowship with and the others don't, for example). maybe i am being too cautious or maybe in the end the benefits outweigh the risks.
  • Explain the rules of the gradient in advance
Changed my mind3
  • a little more time
  • I had some conversations about belief geometry that were very interesting and made me think that it might be even more promising a research direction than I originally thought.
  • After today the ambitious approach to AI safety grounded by theoretical advancements seemed a bit more feasible. Despite just having a relatively short intro to the material, classmates were able to come up with interesting questions that had far-reaching implications for AI safety, and I was a witness to that. On the flip side, I could also see where some research agendas while interesting had a very difficult time connecting with AI safety.
Project proposal day · Fri 30 Oct Fri 30 Oct, leading 4 responses · energy 4.3/5
Overall 0.00 as expected
Content 0.00 as expected
Clarity 0.00 as expected
Motivation 0.00 as expected
Pace 0.00 about right
Difficulty 0.00 about right
Novelty +0.50 mostly new
Read what they wrote4
Worked well1
  • David's presentation on the AI Safety funding landscape was helpful.
Would improve1
  • Some feedback on the proposals would be nice.
Changed my mind1
  • There seem to be more promising directions in data attribution than I previously thought.
Other thoughts1
  • A fifth week on doing a research sprint can be pretty interesting!

The day

10:00 Lecture starts
12:30 Lunch at Mox
13:30 Afternoon session
17:10 Feedback form
17:30 Day ends. Dinner

Resources

Elsewhere