Submitting your AI alignment project

By Adam Jones (Published on May 23, 2024)

If you're participating on the March 2024 Alignment course you'll have your final facilitated session next week. After this you will:

Submit your project here by 10th June. Put the final touches on your project, and submit it to us for judging! We recommend you spend the majority of your efforts after session 11 on creating the final deliverable and communicating what you've done. You should produce some public product as a result of your project such as:

a website (e.g. on GitHub Pages, Notion or Squarespace)
a video (e.g. on YouTube)
a public blog post (e.g. on Ghost or Substack)
some combination of the above
something else entirely

(Optional) Prepare a 3-5 minute presentation of your work. You don’t have to have a slide deck, but you should be able to explain your project to other participants.

Attend the project closing event on 19th June. You’ll present your work to small breakouts with other participants. We’ll also announce the winners of the project competition, award prizes, and close out the course.

Tips on creating an excellent final product

Explain what you managed to achieve, and how this contributes to AI safety. A common mistake is to write about the work you did, rather than what you achieved. Remember that negative results (i.e. when you discovered something didn’t work) are also good to publish!

Target a broad technical audience. Your product should be easy to understand by others - aim for a smart high schooler with some machine learning background. This doesn’t mean you need to dumb down your project - but it does mean you should:

use plain language
define key terms, or link to good definitions of them
structure writing with headers, topic sentences and lists

Ask for feedback. Consider reaching out to connections you made in your learning and project cohorts, or from course networking events. A great tactic is to ask them to explain what they think your project is after reading - this often highlights where your product isn't as clear as you thought it might be.