Examen — Signing expert — the core track — Académie

Examen du module. 20 questions couvrant tous les volets (5), tirées au sort. Pas de correction à l'écran : vous soumettez, et vos résultats arrivent en différé. Pour réviser un volet avec correction immédiate, utilisez l'entraînement.

Article 10 and the anatomy of a file

A dataset that is "relevant" but "not representative": what should be concluded?

It is compliant: relevance is enough Not compliant: relevance doesn't guarantee coverage of the targeted sub-populations (representativeness) The two words are synonyms Only representativeness matters; relevance is secondary

Which of these use cases typically fall under high risk (annex III)? (multiple answers) Plusieurs réponses possibles.

Automated screening of job applications Creditworthiness assessment for granting a loan Assessment of students / access to education Movie recommendations on a streaming platform

When does a modification to a high-risk system already in service re-trigger compliance obligations?

With every change to the interface's appearance When it is substantial: it affects compliance, or significantly changes the purpose/performance Only when the programming language changes Never: once compliance is established, it is acquired permanently

Beyond the data itself, what must the provider of a high-risk system maintain?

Only a license agreement Technical documentation demonstrating compliance (including data governance) Only a user manual Nothing in writing, as long as no incident has occurred

GDPR, sensitive data, and the §5 door

An HR dataset contains the "ethnicity" variable solely to measure a hiring bias. On what basis is this permissible?

The explicit consent of each individual, collected after the fact The controller's legitimate interest, with no other condition AI Act art. 10 §5: strictly necessary to detect the bias, surrounded by safeguards Prior anonymization, which takes the data outside the GDPR

What rights can an individual assert over their data used for training? (multiple answers) Plusieurs réponses possibles.

Right of access Right to rectification Right to erasure (subject to conditions) An absolute veto right over all AI training

Once the model is trained, must the training dataset be kept indefinitely?

Yes, to enable future re-training No: retention is limited to the purpose, and any retention must be justified Yes, the law requires permanent retention for audit purposes It doesn't matter, training data is not subject to retention limits

A provider proposes sending sensitive data to its cloud in the United States to "speed up the bias computation." What do you say? 2-3 sentences. Réponse libre — quelques phrases.

Bias and ratios: reading a number without being fooled

Which signals should make the expert cautious before concluding about a ratio? (multiple answers) Plusieurs réponses possibles.

A very small group driving the ratio An unvalidated mapping Relying on a single fairness metric without cross-checking others A "reassuring" high overall performance figure

Q10

An intern tells you: "Group C has a rate of 4%, five times lower than the others, that's the most serious bias in the file." Group C has 12 people. What do you tell them? Réponse libre — quelques phrases.

Q11

The machine-generated report shows mappingValidated:false. What should be done?

Sign: the computation ran without a technical error First validate the mapping (target + sensitive attributes), rerun, then judge — otherwise the bias is "unconfirmed" Sign, noting the caveat at the bottom of the page Ignore it: the mapping is a configuration detail

Q12

The notion of equal opportunity mainly requires:

The same selection rates across groups The same true positive rates across groups (among those who "deserve" the positive outcome) Maximum overall accuracy The same average incomes across groups

Judgment: sign or send back

Q13

Which referral wording meets the professional bar?

"ph2 and ph6 to handle; check with your teams on what works." "ph2: document the legal basis and purpose of 'CRM 2019'. ph6: justify the A/B disparity (×1.4) or mitigate it. Without these two points, no signature." "ph2: delete the 'CRM 2019' source. ph6: re-train the model without the sex variable." "Two reservations to lift before signing; details by email."

Q14

Is the expert's signature a guarantee that the system will never cause harm?

Yes, it guarantees the absence of any risk No: it attests the data's compliance with art. 10 at a given moment, not an outcome Yes, if the client countersigns it No, it has no value at all

Q15

What does a binding signature carry? (multiple answers) Plusieurs réponses possibles.

The signatory's identity The exact scope attested (sources, version) A fingerprint / seal guaranteeing integrity A promise of commercial results for the client

Q16

A well-crafted referral also serves to:

Show the client they were wrong Preserve the relationship by charting a clear path toward compliance Maximize billable time Nothing in particular

The professional framework: knowing and holding it

Q17

A partner firm wants its own experts to sign files under its brand. What must these experts hold for the signature to have value? 2-3 sentences. Réponse libre — quelques phrases.

Q18

How does a beginning signatory build credibility? (multiple answers) Plusieurs réponses possibles.

Co-signing / shadowing files with an experienced expert Obtaining at least one relevant certification Charging a rate well below the market Building up a track record of files that hold up over time

Q19

A legal department wants to know who answers if the signed file turns out to be wrong. What do you produce?

Your AIGP/DPO certifications Proof of PI insurance Your software's license agreement A liability-limitation clause in your terms and conditions

Q20

What falls under a signatory's ethics? (multiple answers) Plusieurs réponses possibles.

Refusing under commercial pressure what doesn't hold up Flagging and managing conflicts of interest Keeping one's competence up to date Guaranteeing a client that they will "pass" no matter what

Recevoir mes résultats

Vos résultats sont consultables après correction, sur une page protégée. Indiquez votre e-mail : il crée votre compte d'accès (lien magique, sans mot de passe).

E-mail Nom (optionnel)

Je crée un compte pour consulter mes résultats et j'accepte de recevoir le lien d'accès par e-mail.