⚖️ Train your recruitment AI… and watch the bias emerge
You are there HR-robot of a company. Your AI must choose the best candidates among the Rounds 🔵 and the Stars ⭐. Two peoples strictly as competent as each other — only the individual competence (the green bar) should count.
💡 How the AI learns here: we show him the past hires of the company (training data). She deduces on her own what a “good candidate” is. It’s up to you to choose what training data to give it…
Step 1 — Choose training data
Step 2 — Train, then test on 8 new candidates
🗣️ Questions for the class: Has the AI changed between the two trials? (No!) What has changed? (Data.) Who chooses data in real life? How to correct: various data ✔️ tests on all groups ✔️ varied teams ✔️ audits ✔️
🔍 Privacy Checker
Pick up (mentally or actually) your phone, and check off what is true for you . No response leaves this page: everything happens in your browser, even without internet.
🎯 Objective: not fear — lucidity . We observe, we understand, we adjust. Each box checked adds “exposure” points.
🛠️ Your 3 missions to regain control
🛡️ GDPR reminder (General Data Protection Regulation): in Europe, you have the right to see your data, to make them to correct and make them to erase . Minors have reinforced protection. The French gendarme: CNIL .
🃏 Dilemma cards — vote, compare, debate
Real or realistic situations. There is no hidden right answer : there are arguments, values and possible guarantees. Take turns voting on this screen, then view the class results — and debate!
💬 To start the debate: