1. You're fact-checking a viral social media post that includes an AI-generated image of a public official appearing to sign a controversial document. Which TWO layers of the verification stack are most critical to apply, and in which order?
Correct. Layer 1 asks whether independent credible sources confirm the depicted event — and for an image, this means checking whether the event it supposedly shows was reported anywhere with journalistic verification. Layer 4 then asks who created this and what they gain from it being believed.
For visual content, Layer 1 is critical first: does independent reporting confirm the event depicted? Then Layer 4: who benefits from this image spreading? Together these two layers address the most common vectors for AI-generated political visual misinformation.
2. Which type of AI task tends to have the HIGHEST reliability — meaning errors are most likely to be caught immediately?
Correct. Code either runs or it doesn't — the result is immediately verifiable. Other types of claims require external knowledge to evaluate, which is why they're higher risk.
Code testing gives immediate feedback — it works or it doesn't. Current events, legal interpretation, and historical claims all require external knowledge to verify, making errors harder to catch in the moment.
3. What is "model collapse" and how does the increasing volume of AI-generated internet content contribute to it?
Correct. As AI-generated content floods training corpora, successive models train on outputs of previous models, amplifying their patterns and losing connection to the diversity of genuine human knowledge and expression.
Model collapse is the long-term feedback loop risk across model generations. AI trains on AI output, which inherited biases from the AI that generated it — compounding across generations.
4. What term describes when an AI generates plausible-sounding but completely false information, stated confidently?
Correct. AI hallucination means generating false information stated as if it were true — like lawyer Steven Schwartz's invented court cases.
The term is hallucination — generating plausible-sounding false information confidently, as if it were fact.
5. Which of the following would be the BEST application of the "What Would Make This Wrong?" habit?
Correct. The habit requires actively generating potential failure modes for the specific claim — not just seeking alternative opinions. Thinking about what methodology would produce a wrong number, or what contradicting data would look like, is genuine falsification thinking.
The "What Would Make This Wrong?" habit is about actively generating specific failure conditions for the specific claim — not outsourcing disagreement to AI or social media. You have to think: what data, methodology, or logic would prove this claim wrong? That active search for vulnerability is what the habit trains.
6. The 1997 Skitka, Mosier, and Burdick study found that automation bias affects trained experts, not just beginners. What is the most important implication of this for how we design AI systems?
Correct. If even trained experts exhibit automation bias, then relying on individual willpower or expertise to overcome it is insufficient. The implication is that systems must be designed with structural interrupts — alerts, mandatory review steps, override requirements — that don't depend on users recognizing their own bias in the moment.
The key implication is design: if experts are not immune, then individual training cannot be the only solution. Systems need structural safeguards built in — not just better-trained users relying on willpower to override a documented cognitive tendency.
7. Attorney Steven Schwartz's 2023 case is significant primarily because it demonstrated which problem?
Correct.
The case showed that AI hallucinations look exactly like real information — making them extremely difficult for even trained professionals to detect.
8. According to the lesson, why did traditional media literacy skills break down in the face of undisclosed AI content from Sports Illustrated?
Correct. Fake author profiles, use of a trusted brand, and no disclosure of AI authorship each defeat one of the three traditional verification tools — simultaneously.
Each traditional skill relies on accurate source information. Fake authors defeat credential-checking. The trusted brand covers the publication check. And there is no human "who" to investigate. Undisclosed AI removes the foundation all three tools stand on.
9. The Air Canada chatbot case (2024) was legally significant because it held the company responsible for an AI error even though the AI "hallucinated" without intent. What legal standard did the tribunal apply?
Correct. "Reasonable care" is the key standard. It doesn't require intent or perfect accuracy — it requires that companies make reasonable efforts to ensure accuracy before deploying AI to customers.
The standard was "reasonable care" — a middle ground between strict liability and requiring intent. Companies must make reasonable efforts to ensure their AI is accurate. Intent is irrelevant.
10. The Optum healthcare algorithm discriminated against Black patients without using race as a variable. What made this possible?
Correct. Proxy variables can encode historical inequality indirectly — no explicit racial data required. This is one of the most important concepts in AI bias.
The mechanism was a proxy variable: spending standing in for need. The algorithm never "saw" race — it saw the downstream effects of systemic inequality on spending patterns.
11. What was the most significant lesson of the Steven Schwartz legal brief case?
Correct.
The key lesson is about AI confidence not tracking accuracy — the same confident tone applies to fabricated content as to real content.
12. Jonathan Turley was falsely named in an AI-generated list of law professors accused of harassment. The AI also cited a Washington Post article that didn't exist. Which protocol step would catch this specific type of error?
Correct. Step 3 — going to the primary source — would reveal that no such Washington Post article exists. A fabricated citation fails at the first contact with source verification.
Step 3 specifically catches fabricated citations: you go to find the cited source and it doesn't exist. That's the direct catch for this type of error.
13. Which of the following is an example of the COMPLACENCY form of automation bias (not over-reliance)?
Correct. Complacency is reducing your own monitoring because the machine is present — not deferring to a specific machine recommendation. The copy editor isn't following a wrong suggestion; they're simply checking less because they expect the machine to catch errors. The other examples are over-reliance.
Complacency is reducing your own effort because the machine is there. Over-reliance is actively following a machine recommendation against your own judgment. The copy editor stopping their own proofreading is complacency. The other examples all involve actively choosing the machine's wrong recommendation over known-better information.
14. What does the "temperature" setting in a language model control?
Correct. Temperature is the technical control for output variability — it's one reason why the same prompt can get different answers on different runs.
Temperature controls output variability — how much randomness is built into the model's word selection. Higher temperature means more varied, less predictable outputs.
15. Across all four lessons of this module, what is the single most consistent principle underlying every verification technique?
Correct. The module explicitly states this as its core orientation: not blanket trust, not blanket distrust, but systematic verification based on knowing AI's specific failure modes. That's the foundation everything else is built on.
The module explicitly rejects both "always trust" and "always distrust." The core principle is systematic verification based on knowing the specific, predictable ways AI fails — fabricated citations, caveat-stripping, temporal errors, sensory realism in deepfakes. Knowing the failure modes is how you design the checks.
16. What does "confirmation bias" mean in the context of the verification stack, and why does it make Layer 4 particularly difficult?
Correct. Confirmation bias is a human cognitive pattern that makes us check less when we want something to be true. Layer 4's "who benefits" question is hardest to ask about content we already agree with — which is exactly when it's most needed.
Confirmation bias is the human tendency to scrutinize information we dislike while accepting information we like without equivalent rigor. Layer 4 requires overriding this — applying the same question to comfortable information as to uncomfortable information. That's the psychological difficulty.
17. Which type of AI request typically produces the HIGHEST hallucination risk?
Correct. Exact quotes from minor figures represent sparse training data and high specificity — exactly the combination that produces hallucination. AI will generate a plausible-sounding quote rather than admitting it doesn't know.
Hallucination risk spikes with specificity and sparse training data. Exact quotes from minor public figures are both specific and obscure — the AI has to invent something plausible. Creative and general conceptual requests are far lower risk.
18. How does motivated reasoning differ from narrative fit, and why do they combine to make AI misinformation particularly hard to catch?
Correct. They are related but distinct: narrative fit lowers your guard because the claim fits a familiar story; motivated reasoning lowers it further because you want the claim to be true. When both apply simultaneously, the verification threshold drops dramatically — and AI can produce content that triggers both.
They're distinct but compounding. Narrative fit is about story-pattern matching — the claim fits what you expected. Motivated reasoning is about desired outcomes — you want the claim to be true. When both apply, the verification threshold drops the lowest, and that's exactly where AI-generated plausible misinformation slips through.
19. Which of the following claims is LEAST likely to require a currency check?
Correct. 1989 is a historical fact that doesn't change — no currency check needed. The others are all time-sensitive claims where the world continues to evolve past any AI training cutoff.
Historical dates are stable facts — they don't change. Market caps, treatment protocols, and EU membership can all change over time and benefit from a currency check.
20. A researcher using AI to summarize literature for a climate policy brief receives a confident AI summary of a 2021 IPCC report. They notice the AI describes a "95% scientific consensus" figure. What should they do before including this in the brief?
Correct. This scenario requires both source verification (does the IPCC report contain this figure?) and caveat-checking (does the AI's version of the figure drop any qualifications present in the original?). Both layers apply to a real-source situation.
Source authority doesn't eliminate verification needs — the IPCC being credible doesn't mean the AI accurately summarized what the IPCC said. Both source verification (Layer 1) and caveat-checking (Layer 2) apply here.