Megan Kinniment

Here are some examples of projects I have been interested in, but I may be interested in other projects by the time this cohort starts:

• Improve METR's general agent performance using SFT / RL / and scaffolding. (Want to rule out a big capabilities jump with a small amount of effort). 
• Models seem quite bad at judging the quality of their own solutions. Is this a capabilities bottleneck? Is so, can we shortcut running our full task suite by creating a benchmark that just requires models to judge what score an agent trajectory would have gotten?
• How can we convert model X% time horizons into 'expected speedup for person doing X task'?
• How can we model the performance of agents on multistep tasks. (Models like e.g. toby ord's constand hazard rate model seem interesting - but don't fit with our data in some ways)
• Building agentic scheming evals, e.g. where models can choose to sabotage products, and evade or disrupt monitoring. 
• What fundamental weaknesses are holding models back from being dangerous, and how can we test those?

I'll meet with scholars 2x/week each. I'll also be generally available async and potentially for code review. 

Capabilities:

• Measuring AI Ability to Complete Long Tasks
• Follow-up

​

Monitoring:

• SHADE-Arena: Evaluating sabotage and monitoring in LLM agents
• When Chain of Thought is Necessary, Language Models Struggle to Evade Monitors
• CoT May Be Highly Informative Despite “Unfaithfulness”
• Claude, GPT, and Gemini All Struggle to Evade Monitors

Various profiles could be a good fit.

​

Wanted: 

• Enjoys making progress quickly, some 'productive impatience'
• Proactivity and agency (e.g. to unblock yourself, and to ask for help when you need it)
• Comfort doing things you haven't done before (or that nobody has done before!)
• Basic coding skills. (E.g. git, python, bash, know what uv is)
• Able to write reasonably clean code that other researchers can build on
• Prior experience doing research of some kind (broadly defined, could be independent)
• Ok with, or excited to, read some agent transcripts
• Interest in or curious about the models
• Interest in understanding METR as an organization, and how can we better achieve our goals. 
• Enthusiasm is always a plus!

​

Some of the following would be great but not essential: 

• Prior experience making blackbox evaluations.
• Prior experience doing research of some kind (broadly defined, could be independent).
• Prior experience with white-box evaluation techniques. 
• Prior experience creating model organisms. 
• Cybersecurity experience

​

Can independently find collaboraters, but not required