Overview

  • The linked document contains a collection of AI Forecasting research ideas, prepared by some Epoch employees in a personal capacity
  • We think that these are interesting and valuable projects that research interns or students could look into, though they may vary in difficulty (depending on your background/experience)
  • This is the result of a quick brainstorming and curation, rather than a thorough deliberative process. We encourage a critical outlook when reading them
  • You may also be interested in these other forecasting research ideas suggested by Jaime Sevilla 
  • You can use Epoch’s database as a resource for finding notable machine learning papers with parameter, compute, and dataset sizes. On Epoch website you can also find our past researcha tool for visualising the dataset, and some other tools like this compute calculator.
  • Please feel free to contact us for clarification about these questions!
New Comment
2 comments, sorted by Click to highlight new comments since:

> Re: "Extrapolating GPT-N performance" and "Revisiting ‘Is AI Progress Impossible To Predict?’" sections of google doc


Read Section 6 titled "The Limit of the Predictability of Scaling Behavior" of "Broken Neural Scaling Laws" paper: 
https://arxiv.org/abs/2210.14891

Have ya’ll ever considered doing a robust form of strategic foresight in order to have plans for scenarios that fits into a “cone of plausibility”? I think we can learn stuff from forecasting, but I also think the foresight approach is underrated if done well. The main point is that you have thought-through as many plausible scenarios as possible and how everything interacts with one another (technical, socio-economic, geo-political, etc.) and have plans for a variety of scenarios. Even if not of the scenarios happen, I expect you will learn a lot and be much more ready for the scenario that does.

I’m not sure if it’s something you’ve looked into. But if you haven’t, here’s a blog post I wrote that gives a little bit of an overview: https://medium.com/@thibo.jacques/helping-organizations-survive-disasters-and-potentially-avoid-them-altogether-df9a4e835a90

I don’t really do that type of stuff anymore, but would be willing to chat if you have questions.