Seminar 1: Defend Your Analysis

Friday 10:00–12:00 CEST · live on Zoom · mandatory.

The Friday seminar is where you defend the work you did in Wednesday’s computer lab. It is not a lecture and it is not a paper-reading club — you present your own analysis, and the room questions it. This page explains how every seminar in the course works; the format below is the same in weeks 1–6.

What the seminar is (and why it replaced the journal club)

In previous years this slot was a journal club where we read someone else’s paper. This year the course trains you to be a forward-deployed scientist — to take someone else’s problem, direct an AI agent to solve it, and own every number that comes back. Owning a result means being able to stand behind it out loud, so the seminar is now your own work, defended.

Think of it as the weekly checkpoint for the skill the labs build: not “did the code run” but did you interview well, direct clearly, and verify honestly.

How it runs

• Presenters are drawn at random. Each week we pick 7–10 students live. Everyone must arrive ready — there is no volunteering-only and no opting out.
• 7 minutes to present + 3 minutes of discussion, per person. Keep to time; we will.
• You present the work from this week’s lab — the problem you were handed, what you built, and how you checked it.
• The rest of the room (and the teaching team) asks questions. Being able to answer them is the exam.

Being drawn with nothing prepared is a fail for that seminar. The random draw is the whole point: it means everyone spends Thursday thinking critically about their own work, not just the volunteers.

What to present — four things

You have 7 minutes. Don’t narrate the whole afternoon; make these four points land.

1. The problem — the question behind the question. What did the data owner actually want? What decision changes when they know the answer? Show that you found the real question, not the vague one they opened with.
2. What you built. How you translated the interview into a brief (AGENTS.md / spec.md) and directed the analyst agent. One or two slides — the shape of the pipeline, not every command.
3. A critical read of your own method. Where could it be wrong? What did you check — which controls did you run (shuffled labels, dumb baseline, batch/split check)? And, crucially, what did you refuse to claim?
4. What the literature says. How have others approached this kind of problem? One or two references is enough — it shows you placed your result in context rather than in a vacuum.

The single best thing you can say in a DDLS seminar is “here is the result, here is the control that could have killed it, and here is what I still would not claim.” That sentence is what we are listening for.

How you’re assessed

Seminars are graded pass/fail on preparation and critical engagement — not on getting a “good” result. A modest finding, honestly interrogated, passes easily. A polished number with no controls and no caveats does not.

We are specifically listening for:

• Did you find the question behind the question?
• Where did you refuse what the agent handed you?
• What did you check, and what did you take on faith?

For Master’s students, the course grade (A–F) comes from the final oral defence, not the weekly seminars — but the seminars are where you rehearse exactly that skill. Treat every Friday as a low-stakes run at the exam.

Attendance

Labs and seminars are the mandatory, live core of the course. You may miss one mandatory session in total — a lab or a seminar, not one of each. If you must miss the seminar, email ddls-course@scilifelab.se before the session, not after.

How to prepare (Thursday, ~30 minutes)

You already did the work in the lab; preparing to defend it is quick:

• Pull your three artefacts from the lab: the interview transcript (Agent A), the analysis transcript (Agent B), and your short report with AGENTS.md / spec.md. Pi saves each run as a .jsonl file under ~/.pi/agent/sessions/ — if you launched Pi more than once, collect them all; together they are your Agent-B transcript.
• Draft 4–5 slides around the four points above. Slides optional but recommended — you can even have your agent draft them from your report, then fix them.
• Rehearse the 7 minutes once, out loud. If you run long, cut the “what I built” detail first — the problem and the verification matter most.
• Look up one or two references on how your problem is usually approached, for point 4. Pi has no web search, so use your own browser — or have Pi fetch a citation from its shell (e.g. curl 'https://api.crossref.org/works?query=YOUR+TOPIC'). Never let the agent invent a DOI: check that every reference actually resolves before you cite it.

This week’s lab sets up everything you’ll present. If you haven’t done it yet, start there:

• Go to Computer Lab 1