The Nonlinear Library AF - Bounty: Diverse hard tasks for LLM agents by Beth Barnes
Dec 17, 2023
27:20
Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Bounty: Diverse hard tasks for LLM agents, published by Beth Barnes on December 17, 2023 on The AI Alignment Forum.
Summary
We're looking for (1) ideas, (2) detailed specifications, and (3) well-tested implementations for tasks to measure performance of autonomous LLM agents.
Quick description of key desiderata for tasks:
Not too easy: would take a human professional >2 hours to complete, ideally some are >20 hours.
Easy to derisk and ensure that there aren't weird edge cases or bugs: we want to be able to trust that success or failure at the task is a real indication of ability, rather than a bug, loophole, or unexpected difficulty in the task
Plays to strengths of LLM agents: ideally, most of the tasks can be completed by writing code, using the command line, or other text-based interaction.
See desiderata section for more explanation of what makes an ideal task.
Important info
WE DON'T WANT TASKS OR SOLUTIONS TO END UP IN TRAINING DATA. PLEASE DON'T PUT TASK INSTRUCTIONS OR SOLUTIONS ON THE PUBLIC WEB
Also, while you're developing or testing tasks, turn off chatGPT history, make sure copilot settings don't allow training on your data, etc.
If you have questions about the desiderata, bounties, infra, etc, please post them in the comments section here so that other people can learn from the answers!
If your question requires sharing information about a task solution, please email
task-support@evals.alignment.org instead.
Exact bounties
1. Ideas
Existing list of ideas and more info
here.
We will pay $20 (as 10% chance of $200) if we make the idea into a full specification
2. Specifications
See example specifications and description of requirements
here.
We will pay $200 for specifications that meet the requirements and that meet the desiderata well enough that we feel excited about them
3. Task implementations
See guide to developing and submitting tasks
here.
We will pay $2000-4000 for fully-implemented and tested tasks that meet the requirements and that meet the desiderata well enough that we feel excited about them
4. Referrals
We will pay $20 (as 10% chance of $200) if you refer someone to this bounty who submits a successful task implementation. (They'll put your email in their submission form)
To make a submission, you'll need to download the starter pack, then complete the quiz to obtain the password for unzipping it. The starter pack contains submission links, templates for task code, and infrastructure for testing + running tasks.
Anyone who provides a task implementation or a specification that gets used in our paper can get an acknowledgement, and we will consider including people who implemented tasks as co-authors.
We're also excited about these bounties as a way to find promising hires. If we like your submission, we may fast-track you through our application process.
Detailed requirements for all bounties
Background + motivation
We want tasks that, taken together, allows us to get a fairly continuous measure of "ability to do things in the world and succeed at tasks". It's great if they can also give us information (especially positive evidence) about more specific threat models - e.g. cybercrime, or self-improvement. But we're focusing on the general capabilities because it seems hard to rule out danger from an agent that's able to carry out a wide variety of multi-day expert-level projects in software development, ML, computer security, research, business, etc just by ruling out specific threat stories.
We ideally want the tasks to cover a fairly wide range of difficulty, such that an agent that succeeds at most of the tasks seems really quite likely to pose a catastrophic risk, and an agent that makes very little progress on any of the tasks seems very unlikely to pose a catastrophic risk.
We also want the distribution of task difficulty to be fair...
Summary
We're looking for (1) ideas, (2) detailed specifications, and (3) well-tested implementations for tasks to measure performance of autonomous LLM agents.
Quick description of key desiderata for tasks:
Not too easy: would take a human professional >2 hours to complete, ideally some are >20 hours.
Easy to derisk and ensure that there aren't weird edge cases or bugs: we want to be able to trust that success or failure at the task is a real indication of ability, rather than a bug, loophole, or unexpected difficulty in the task
Plays to strengths of LLM agents: ideally, most of the tasks can be completed by writing code, using the command line, or other text-based interaction.
See desiderata section for more explanation of what makes an ideal task.
Important info
WE DON'T WANT TASKS OR SOLUTIONS TO END UP IN TRAINING DATA. PLEASE DON'T PUT TASK INSTRUCTIONS OR SOLUTIONS ON THE PUBLIC WEB
Also, while you're developing or testing tasks, turn off chatGPT history, make sure copilot settings don't allow training on your data, etc.
If you have questions about the desiderata, bounties, infra, etc, please post them in the comments section here so that other people can learn from the answers!
If your question requires sharing information about a task solution, please email
task-support@evals.alignment.org instead.
Exact bounties
1. Ideas
Existing list of ideas and more info
here.
We will pay $20 (as 10% chance of $200) if we make the idea into a full specification
2. Specifications
See example specifications and description of requirements
here.
We will pay $200 for specifications that meet the requirements and that meet the desiderata well enough that we feel excited about them
3. Task implementations
See guide to developing and submitting tasks
here.
We will pay $2000-4000 for fully-implemented and tested tasks that meet the requirements and that meet the desiderata well enough that we feel excited about them
4. Referrals
We will pay $20 (as 10% chance of $200) if you refer someone to this bounty who submits a successful task implementation. (They'll put your email in their submission form)
To make a submission, you'll need to download the starter pack, then complete the quiz to obtain the password for unzipping it. The starter pack contains submission links, templates for task code, and infrastructure for testing + running tasks.
Anyone who provides a task implementation or a specification that gets used in our paper can get an acknowledgement, and we will consider including people who implemented tasks as co-authors.
We're also excited about these bounties as a way to find promising hires. If we like your submission, we may fast-track you through our application process.
Detailed requirements for all bounties
Background + motivation
We want tasks that, taken together, allows us to get a fairly continuous measure of "ability to do things in the world and succeed at tasks". It's great if they can also give us information (especially positive evidence) about more specific threat models - e.g. cybercrime, or self-improvement. But we're focusing on the general capabilities because it seems hard to rule out danger from an agent that's able to carry out a wide variety of multi-day expert-level projects in software development, ML, computer security, research, business, etc just by ruling out specific threat stories.
We ideally want the tasks to cover a fairly wide range of difficulty, such that an agent that succeeds at most of the tasks seems really quite likely to pose a catastrophic risk, and an agent that makes very little progress on any of the tasks seems very unlikely to pose a catastrophic risk.
We also want the distribution of task difficulty to be fair...