Ought will host a factored cognition “Lab Meeting” on Friday September 16 from 9:30AM - 10:30AM PT.
We'll share the progress we've made using language models to decompose reasoning tasks into subtasks that are easier to perform and evaluate. This is part of our work on supervising process, not outcomes. It’s easier for us to show you than to tell you about it in a post (though written updates will hopefully follow).
Then, we'll cover outstanding research directions we see and plan to work on, many almost shovel-ready. If the alignment community can parallelize this work across different alignment research teams, we can make progress faster. We'd love to coordinate with other alignment researchers thinking about task decomposition, process supervision, factored cognition, and IDA-like approaches (where efficient to do so). We want to save you time and mistakes if we can!
There will be more to discuss than we can fit into an hour. We’ll get to what we can and consider making this a regular meeting if there’s appetite (likely with more sharing from other researchers)!
You should attend if:
You can register for the Lab Meeting here. Email firstname.lastname@example.org if you have any questions!
The meeting will be recorded & shared.
The video from the factored cognition lab meeting is up:
Ought cofounders Andreas and Jungwon describe the need for process-based machine learning systems. They explain Ought's recent work decomposing questions to evaluate the strength of findings in randomized controlled trials. They walk through ICE, a beta tool used to chain language model calls together. Lastly, they walk through concrete research directions and how others can contribute.
00:00 - 2:00 Opening remarks2:00 - 2:30 Agenda2:30 - 9:50 The problem with end-to-end machine learning for reasoning tasks9:50 - 15:15 Recent progress | Evaluating the strength of evidence in randomized controlled trials trials15:15 - 17:35 Recent progress | Intro to ICE, the Interactive Composition Explorer17:35 - 21:17 ICE | Answer by amplification21:17 - 22:50 ICE | Answer by computation22:50 - 31:50 ICE | Decomposing questions about placebo31:50 - 37:25 Accuracy and comparison to baselines37:25 - 39:10 Outstanding research directions39:10 - 40:52 Getting started in ICE & The Factored Cognition Primer40:52 - 43:26 Outstanding research directions43:26 - 45:02 How to contribute without coding in Python45:02 - 45:55 Summary45:55 - 1:13:06 Q&A
The Q&A had lots of good questions.