Plan for Tuesday, 31 March
Each project team will be expected to give a brief pitch (no more than two minutes) for your project.
If you would like to use slides for this or show a website, send me a link before 9pm on Monday (March 30).
You should not have more than 4 slides (not counting a title slide), and no slide should contain more than 10 words, unless you have a very good reason for violating either of these guidelines, in which case, you should explain it in your email with the link.
If you missed the class where I talked about presentations, please review Class 14.
For your project pitches on Tuesday, I would recommend (unless your team thinks something different is appropriate for your project):
-
Title slide with your project title and your names.
-
Clear statement of the problem, story that motivates your idea
-
Explain/show what you are actually doing
-
Describe the most interesting thing you have found or done so far.
-
(If not obvious from 1–3) What you are planning to do for the rest of the project.
It is not necessary for everyone on the team to speak (and probably not advisable to have more than 2 people speak in a two-minute presentation), but everyone on the team can help answer questions.
Reading for Thursday, 2 April
- Ilia Shumailov, Zakhar Shumaylov, Yiren Zhao, Nicolas Papernot, Ross Anderson and Yarin Gal. AI models collapse when trained on recursively generated data. Nature, July 2024.
Don’t worry if you don’t have the mathematical background to understand the technical parts of the “Theoretical intuition”, but you should get an understanding of what they mean at a high level.
The paper below is optional, but I’m expecting the lead team to summarize its key points:
- Rylan Schaeffer, Joshua Kazdan, Alvan Caleb Arulandu, and Sanmi Koyejo. Position: Model collapse does not mean what you think, 2025.