Polimi PSSD · AI for Social Transformation Design

Working with AI:
field notes

The short version of what I’ve learned helping students use generative AI well. Tips and insights — not prompts. Play first, then purpose.

A few truths

AI is (probably) inevitable. It’s being baked into Office, Google Docs, Canva, Miro — it will be hard to escape.
It keeps accelerating. It might “plateau” soon — I wouldn’t bet on it.
Most people are afraid of it — and not very good at it. The fear prevents the play that builds the skill.
It raises the ceiling. Everyone can code now. It’s mostly an attitude.
New learning goals. Authenticity, curiosity, intrinsic motivation. Experiment as much as you can.

It’s up to you. AI augments you; it isn’t autonomous. Bad results? Assume there’s a better prompt.
Be specific. Vague questions get generic answers.
Play with it. Effective use comes from intuition — and intuition comes from play.
Vibe, don’t engineer. “Prompt engineering” is too left-brained. Vibe with it.
Crap begets crap. A lame reply poisons the ones after it. Edit your input until quality returns.
Quality breeds quality. Prime it with rich context before you ask the real question.
Start fresh. New chat per task — quality degrades over a long thread.
Don’t turn your brain off. Give it your own ideas; it thinks better. And it can’t finish the job — the last 10% is half the work.
Ask it what it’s planning before it does it, so you can steer.

Think critically. Assume anything it says could be false — perspective, not truth. Verify.
Be authentic. Tell it your real purpose. Use it to figure out what you actually want.
Have integrity. Know what you want to say. Don’t cite what you haven’t read.
Own the outcome. Like a camera — you didn’t paint the pixels, but you framed it, and you own it.
AI makes parts, not wholes. Copy the good selections, not the whole.
Edit extensively. Learn what “sounds like AI” means — and cut it.
Start early. It doesn’t make you faster; it makes it easier to start.
Don’t be lazy. Supplement your thinking, don’t replace it. Raise your own standards.

These are the common misconceptions — each one is worth questioning. Usually the truth is more interesting than the fear.

The paradox: it can do 90% of the work — but the last 10% of a project is usually 90% of the work. Don’t wait till the end.