Europe 2031, and what AI actually is
HEC Montréal
qmib.load(...), and know where the file actually isEverything you need to type is on one sheet — how to start Python in VS Codium on your platform, how to make a file and run it, and the four DataFrame moves you will use every week from now on.
Python Cheatsheet · Session 01 (PDF) · in the repository at
01-…/PYTHON-CHEATSHEET.pdf
Print it, or keep it open beside the editor. Nothing on it is examinable; all of it is assumed.
Europe 2031 is useful precisely because it mixes four different kinds of statement — and moves between them without announcing it.
| Evidence | A claim tied to a source and an observed time period |
| Scenario | A coherent imagined future — not a forecast by default |
| Prediction | A dated, testable statement about an outcome |
| AI | A computational system producing outputs from inputs and learned parameters |
Telling these apart, in a text that does not label them, is the skill. It is also the whole of the final paper.
| 00–10 | Opening | One sentence each: the claim that stayed with you |
| 10–25 | Evidence | Locate sources, dates, and the boundary of what is observed |
| 25–45 | Scenario | Identify assumptions; make one claim measurable |
| 45–50 | Pause | |
| 50–70 | AI | Use the local model to question one passage — then verify what it says |
| 70–85 | Conversation | Compare groups; name absent evidence and alternative futures |
| 85–90 | Close | One claim, one uncertainty, one question |
The scenario has three parts, and they do different work:
| Observed | Which statements have dated sources? |
| Constructed | Where do assumptions begin doing the work? |
| Rhetorical | Which framing makes one future feel inevitable? |
The seam between the first two is rarely marked in the text. Find it.
Source stop. What is the source, the date, the unit, and the observed period?
Measurement stop. What variable would make this claim testable?
Alternative stop. What mechanism, or what future, is missing?
Reversal stop. What evidence would change your mind? If nothing would, you are not holding a claim — you are holding a position.
\[\text{prompt tokens} \;\rightarrow\; \text{learned parameters} \;\rightarrow\; \text{next-token scores} \;\rightarrow\; \text{generated text}\]
What it can do. Propose questions, transformations, code, summaries, alternative phrasing.
What it cannot guarantee. Truth. Valid sources. Causal identification. Reproducibility. Responsibility.
Everything in this course that looks like a rule about AI use follows from that second list.
Take one passage from Europe 2031. Ask your local model to state the strongest objection to it. Then verify the objection — is the source real, is the number right, does the mechanism hold?
Whatever you find is the first entry in the habit this course grades: an LLM output you identified as wrong, unverifiable, or misleading, and how you established that.
Ninety seconds per group. Everyone speaks.
| One claim | The narrowest claim the evidence supports |
| One uncertainty | The assumption, or the missing evidence |
| One question | Carry it into Session 02 |
Listen for traceable evidence, not confidence of delivery.
Read the-supervised-learning-problem.md if you would like a head start — optional, and not examinable. Session 02 introduces everything it needs.
The scenario: europe2031.ai · The critical brief: reading-europe-2031.md
MATH60033A · Session 01 practice · Thierry Warin, PhD · HEC Montréal