1. Adversarial examples demonstrate which structural vulnerability of CNN-based vision systems?
Correct. Adversarial examples exist because CNNs optimize statistical pixel-label correlations, not semantic understanding. Perturbations that redirect those correlations — even invisibly — can reliably produce confident misclassifications.
Adversarial vulnerability is structural: CNNs learn statistical pixel-label correlations, not meaning. Perturbations exploit these correlations in ways that are invisible to human perception but redirect the network's output entirely.
2. Chicago's gang database (as of 2018) listed approximately what percentage of Black male Chicagoans over age 16?
Correct. Approximately 32% of Black male Chicagoans over 16 were listed in the gang database — a figure that illustrates how broad and demographically concentrated such systems can become, often without conviction or due process.
The figure was approximately 32% — nearly one in three Black men over 16 in Chicago was listed, most without charges or convictions, with significant collateral consequences for employment and housing.
3. A city bans government use of facial recognition technology in public spaces. Which city did this first, in 2019?
Correct. San Francisco banned government use of facial recognition in 2019, the first major US city to do so.
Not quite. San Francisco was the first major US city to ban government facial recognition, in 2019.
4. Microsoft's retraction of the MS-Celeb-1M dataset was prompted by what finding?
Correct. Researchers investigating MS-Celeb-1M found it included images of private individuals scraped without consent, prompting Microsoft to retract it — illustrating the provenance and consent problems in facial recognition training data.
Not quite. MS-Celeb-1M was retracted after researchers found it included images of private individuals collected without their knowledge or consent — a training data provenance and ethics problem.
5. "Representation bias" in a training dataset means:
Correct. Representation bias is about proportion — some populations are underrepresented, so the model has weaker internal representations for them.
Not quite. Representation bias specifically refers to unequal frequency of different groups in training data — not annotation quality or consent.
6. Amazon's Just Walk Out technology first opened to the public in which city and year?
Correct. Amazon Go opened to the public on January 22, 2018, at 2131 7th Avenue in Seattle.
Amazon Go opened to the public on January 22, 2018, in Seattle.
7. Robert Williams was wrongfully arrested in Detroit in January 2020. How long was he held before the error was acknowledged?
Correct. Williams was held for approximately 30 hours before police acknowledged the facial recognition match was incorrect.
Not quite. Williams was held for about 30 hours. Nijeer Parks spent ten days in jail in a separate NJ case.
8. Illinois' Biometric Information Privacy Act (BIPA) was enacted in:
Correct. BIPA was enacted in 2008, over a decade before most biometric privacy discussions entered mainstream policy debate. Its foresight — and its private right of action — made it the most powerful tool in U.S. biometric privacy litigation.
BIPA was enacted in 2008 — well before facial recognition became a mass-market technology. This foresight is what made it available as a legal tool when Clearview, Facebook, and others later deployed biometric systems at scale.
9. The 2019 Nature Medicine dermatology AI study found its system performed worst on which patient group?
Correct. The training dataset's underrepresentation of darker skin tones directly reduced diagnostic accuracy for those patients.
Not quite. The training images skewed heavily toward lighter-skinned patients, leaving the system less reliable for those with darker skin tones.
10. San Francisco's 2019 facial recognition ban was followed by at least how many other U.S. cities enacting similar measures through 2022?
Correct. Oakland, Somerville, Boston, Portland, and more than a dozen other cities followed San Francisco's lead in banning government use of facial recognition through 2020–2022.
Not quite. More than a dozen U.S. cities enacted similar municipal bans following San Francisco, including Oakland, Somerville, Boston, and Portland.
11. DenseNet's key architectural innovation, which made it useful for deep medical image models, is:
Correct. DenseNet connects each layer to every subsequent layer (not just the next), ensuring gradients can flow through the full 121-layer network without vanishing.
Incorrect. DenseNet's defining feature is that each layer receives concatenated feature maps from all previous layers — this dense connectivity preserves gradient flow across very deep networks.
12. The 2019 NIST Face Recognition Vendor Test found that many commercial algorithms showed false-positive rates for Black women that were how much higher than for white men?
Correct. The NIST study documented false-positive differentials of 10 to 100 times for Black women compared to white men across many commercial algorithms — a disparity large enough to constitute a systematic, not incidental, problem.
The NIST findings were much more severe than minor variation — 10 to 100 times higher false-positive rates for Black women represented one of the most significant algorithmic bias findings in computer vision research.
13. "Data minimization" as a surveillance governance principle means:
Correct. Data minimization is a foundational privacy principle: collect only what you need for a specific purpose, and delete it when that purpose is complete — preventing surveillance data from accumulating into permanent searchable archives.
Data minimization refers to the scope and duration of collection — only what's needed, only for as long as needed. It's about preventing the accumulation of broad surveillance archives beyond what any specific purpose requires.
14. The 2023 revelation about Amazon's Just Walk Out system showed that even advanced deployed AI often requires:
Correct. Over 1,000 workers in India reviewed ambiguous transactions — illustrating human-in-the-loop as a real feature of production AI systems, not a temporary limitation.
The revelation was that over 1,000 human reviewers in India handled transactions the AI couldn't confidently resolve — a real-world example of human-in-the-loop at scale.
15. What data structure does Viola-Jones use to compute rectangular pixel sums in exactly four operations?
Correct. The integral image (summed area table) precomputes cumulative pixel sums so that any rectangular region sum requires only four lookups, making Haar feature computation extremely fast.
The integral image is the data structure that enables constant-time rectangular sum computation, a key speed component of Viola-Jones.
16. The main advantage radar has over camera and lidar in poor weather is:
Correct. Radio waves are largely unaffected by precipitation, and the Doppler frequency shift provides direct velocity measurement — two capabilities cameras and lidar lack.
Radar's key advantage is weather penetration and direct velocity measurement via Doppler. It has low spatial resolution — it cannot read signs or identify object shapes precisely.
17. The NYPD's Demographics Unit, which conducted surveillance of Muslim communities from 2002–2014, produced how many terrorism leads over its entire operation?
Correct. The NYPD's own documentation of the Demographics Unit showed zero terrorism leads generated — while the community impact, including self-censorship and reduced mosque attendance, was documented extensively in subsequent litigation.
The NYPD's own reports acknowledged zero terrorism leads. The program operated for over a decade, surveilling an entire community, with no documented counterterrorism benefit.
18. What did the 2020 New England Journal of Medicine study find about pulse oximeters during the COVID-19 pandemic?
Correct. The calibration bias introduced when devices were designed using lighter-skinned participants produced life-threatening measurement gaps during COVID-19.
Not quite. The study found a roughly threefold disparity in missed low-oxygen readings for Black patients — a direct consequence of non-representative calibration data.
19. A Grad-CAM heatmap that highlights the edge or background of a medical image (rather than the lesion) should lead a radiologist to:
Correct. Research has shown radiologists appropriately lower their trust in AI outputs when Grad-CAM heatmaps highlight irrelevant regions — a key reason interpretability tools improve human-AI collaboration quality.
Incorrect. If Grad-CAM shows the model is focusing on clinically irrelevant areas like image edges, that is a red flag indicating shortcut learning or spurious correlation — the radiologist should reduce trust in that output.
20. Mobileye's REM (Road Experience Management) system collects mapping data primarily from:
Correct. REM turns every equipped vehicle into a passive mapping probe, crowdsourcing HD map data at massive scale — over 8 billion km collected across 40+ countries by 2023.
REM is a crowdsourced system: vehicles equipped with Mobileye chips upload anonymized road geometry data in the background. This scales map coverage far beyond what a dedicated fleet could achieve.