Digital Product Design, Emerging Tech, and Responsible AI
Free AIPMM Certified Digital Product Manager practice — 6 questions on Digital Product Design, Emerging Tech, and Responsible AI, with explanations. No sign-up. Full 12-question mixed test →
A national bank's mobile deposit app must meet accessibility requirements before a public launch in the US. Legal has flagged that litigation risk is high if the app is not accessible to screen reader users, low-vision users, and users with motor impairments. The design team argues for pursuing every WCAG AAA success criterion to eliminate risk entirely. As the digital PM, what is the BEST accessibility target to set for the release?
WCAG AA is the conformance level referenced by ADA-related settlements, Section 508, and most international accessibility regulations, making it the practical and defensible target. AAA includes criteria (e.g., sign language interpretation for all video, no timing) that W3C itself states are not achievable for all content types.
A social platform's content-moderation ML classifier flags posts as 'harmful' or 'safe.' Missing a harmful post (false negative) causes serious user-safety and PR damage; incorrectly flagging a safe post (false positive) causes user annoyance but is easily appealed. As the product manager setting the launch threshold with the ML team, which approach should you prioritize?
When the cost of a false negative (missed harmful content) is much higher than a false positive, the PM should tune the model toward recall on the harmful class and offset the resulting increase in false positives with a lightweight appeals/override path, which is a standard responsible-AI mitigation for asymmetric error costs.
A payments company wants to add fraud detection to a new low-volume B2B invoicing feature. Transaction volume is small (a few hundred per week), the fraud patterns are well understood and stable (e.g., mismatched billing/shipping country plus a first-time vendor), and the compliance team requires that every flagged decision be fully explainable to auditors. Which approach should the PM recommend for the initial release?
When data volume is small, patterns are well understood, and full explainability is a compliance requirement, a rules-based system is the right initial approach: it needs no training data, is fully auditable, and matches the known patterns exactly. ML should be reserved for cases where patterns are complex, data volume is sufficient, and some accuracy-for-explainability tradeoff is acceptable.
Your platform team maintains a public REST API used by dozens of third-party integrators. You need to ship a breaking change to a core resource's response schema. Which API change-management configuration BEST protects existing integrators while allowing the platform to evolve?
Standard API governance practice for breaking changes is explicit versioning plus a published deprecation/sunset policy with advance notice, so integrators can migrate on their own schedule without unplanned outages.
A product team wants to improve a support chatbot by fine-tuning it on two years of real customer support chat transcripts, which include names, account numbers, and other personal data. Before approving this data reuse, what should the PM treat as the MOST critical gating consideration?
Repurposing personal data for a new use (training an AI model) beyond its original collection purpose is a core privacy/purpose-limitation issue. The PM must confirm the legal basis or consent covers this secondary use and drive de-identification/anonymization or re-consent before engineering work proceeds, since this is a compliance blocker, not a technical detail.
A fintech company is launching an AI model that recommends loan approval/denial decisions to loan officers. The model's recommendations directly influence which applicants get credit, a high-stakes and legally regulated outcome. Which product design decision BEST reflects responsible-AI practice for this launch?
For high-stakes, regulated decisions like credit approval, responsible-AI practice calls for human-in-the-loop review with explainable contributing factors and an audit trail of agreement/override, supporting both accountability and compliance with fair-lending and adverse-action disclosure requirements.
Ready for the real thing?
The full course: two full-length practice tests, video lessons for every exam domain, hands-on labs and detailed explanations.
$109.99 $34.99 with code FREETEST33 — valid through September 7.
Get my $34.99 deal →