Cover Image for AI Alignment CRITIQUE-A-THON
Cover Image for AI Alignment CRITIQUE-A-THON
Avatar for AI-Plans
Presented by
AI-Plans
Hosted By
3 Went

AI Alignment CRITIQUE-A-THON

Virtual
Registration
Past Event
Welcome! To join the event, please register below.
About Event

Dissect AI Alignment Plans

AI-Plans.com is hosting a Critique-a-Thon of AI Alignment Plans from May 25 to 26th.

Debate with a partner for and against a Critique applying to an Alignment Plan.

Swap sides the next day.

Conclude with a write-up. (For an example, see The December Winner.)

Get feedback from expert judges, on your write-up.

Judges include:

  • Ramana Kumar, former Senior Researcher at DeepMind

  • Charbel-Raphaël Segerie, head of the AI unit at EffiSciences

  • Dr Peter S Park, MIT postdoc at the Tegmark lab

  • Dr Atticus Geiger, Stanford PhD. Principal Investigator at Pr(Ai)²R Group

Who is this event for?

If you are familiar with the basics of AI Alignment, e.g. you've completed the AI Safety Fundamentals course, or have equivalent or greater knowledge, you'll definitely be able to contribute.

Everyone, regardless of their experience in AI Safety is very welcome to attend the Presentation Evening

Schedule

Stage 1 - First Debate (25th)

We'll split into pairs. Each pair will select an alignment plan that is less than 3 years old. One will make the case for, the other will make the case against, the strength/vulnerability being true and accurate.

Then, the next day, we'll swap sides and finish with a write-up of the proposed strength/vulnerability.

See previous winning critiques for examples:

December Critique-a-Thon

September Critique-a-Thon

August Critique-a-thon

The gold standard to aim for: https://arxiv.org/pdf/2307.15217.pdf

Stage 2 - Discussion & Write-Up (26h)

Change partners and repeat the process from Stage 1.

Submission Deadline: 23:59, May 26th

Presentation Evening (9pm GMT to 11pm GMT, May 26th)

Teams will have a chance to present what they worked on and what they learnt about the alignment plans they chose.

Everyone is welcome to join, even if you didn't take part in the Debate.

The discussions will help refine the critiques and strengthen the arguments. Swapping reduces disadvantages from a lack of prior knowledge on a topic, e.g. if you're discussing an RL section of a paper and your partner has more experience in that area.

If your partner knows a lot more about the topic and makes points you hadn’t thought of, you can use them yourself the next day and see what the counters are - and identify if they’re something to be mentioned in your write-up, which is what will be judged.

Discussions will take place on the Discord.

Join here 👉 https://discord.gg/aGVtu5JyjJ




Avatar for AI-Plans
Presented by
AI-Plans
Hosted By
3 Went