Repository navigation
Discussion: FID-107 — What Makes an AI Safety Research Task a Useful First Contribution? #233
alexchaomander
started this conversation in
Research Calls
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Which task designs help first-time research contributors produce a result another researcher can use within an agreed time and review budget?
This seed call welcomes questions, reading notes, and offers to draft or review a bounded first contribution.
Starter project
Start with one existing task format. Read the research-task template and
the public two-source review issue as an example. For a reproducible dry-run,
choose the FID-106 offline result check or the
two-source synthesis.
Research-task template ·
Example two-source issue ·
Study of mentoring on good first issues
Dry-run the task for 30 minutes. Follow only the written materials first.
Record the first avoidable ambiguity, access problem, or handoff issue. If
you reach the output without one, record that and what made it possible.
Propose one small revision and give an independent reviewer enough detail to
see whether it addresses the problem.
Done means a short record shows the task, the time spent, what you could
produce, and the proposed edit or reason the existing packet suffices. A
reviewer can check the artifact against its completion criteria. A coordinator
can estimate the time to keep the revision current. This individual dry-run
is separate from any comparative study with participants.
Two-source option for FID-045 and FID-107
Read the methods and limitations of Omissive Bias in Religious Representation
and When AI Takes Sides on Questions of Faith.
Using the source-note template, draft a one-page
comparison of the user context, outcome and reference judgment in each study.
Name one rubric decision the evidence could inform and one comparison it cannot
support. For the synthesis, scope this as a 1–2 hour task. For the FID-107
dry-run, stop at the 30-minute mark and record any blocker. Check the existing
source reviews before proposing a duplicate annotation.
Comment with the part you would like to take on, your rough availability, and any review or support you need. The contributor guide explains how to agree a first output. Use the private introduction route for contact details or confidential context.
All reactions