1. Klarna's 2024 AI assistant announcement is notable in the labor market evidence base because:
Correct. Klarna is notable for showing a genuine replacement effect — not just augmentation — at organizational scale. Combined with the company's prior workforce reduction from 7,000 to 3,800, it represents one of the clearest documented cases of AI directly reducing headcount rather than simply increasing output per worker.
The Klarna case is significant because of the combination: AI handling 700-agent volume, with the company having already cut its workforce nearly in half. That pairing documents replacement rather than augmentation — still relatively rare in the AI deployment literature.
2. Bill Gates proposed the robot tax concept primarily to address:
Correct. Gates' 2017 argument was fiscal: if a robot replaces a worker who paid income taxes, and the robot's profits are taxed at lower capital rates, the social services that depend on labor tax revenue will be defunded. A robot tax neutralizes this shift.
Gates' robot tax proposal was about fiscal sustainability — automation shifts income from labor (high tax rates) to capital (lower rates), eroding the tax base that funds social services. The tax would offset this structural revenue shift.
3. The 2023 NEJM AI study found that radiologists using AI assistance achieved what result compared to AI alone?
Correct. The human-AI combination beat both AI alone (by 14%) and humans alone (by 11%) — the empirical case for Tier 2 skills (domain expertise + AI fluency).
Incorrect. The combination achieved 14% lower errors than AI alone and 11% lower than radiologists alone — the combination outperformed either component.
4. What does a "standing joint technology committee" in a union contract do?
Correct. Standing committees address the fundamental problem that multi-year contracts can't anticipate every AI tool — they give unions ongoing engagement rights between bargaining rounds rather than waiting years for the next contract cycle.
Standing joint technology committees solve the temporal mismatch between static contracts and continuous AI deployment — giving unions a mechanism to engage with new tools without waiting for the next contract.
5. Klarna's 2024 AI assistant handling 2.3 million conversations displaced roles primarily because those roles scored near zero on:
Correct. Scripted customer service resolution is low-variability by design — exactly the type of work AI handles most reliably.
Incorrect. Scripted customer service follows predictable paths — low task variability — making it ideal for AI substitution.
6. The Frey-Osborne (2013) estimate that 47% of U.S. jobs were at high automation risk has been widely criticised primarily because it:
Correct. The core methodological critique is that assessing risk at the job-title level masks enormous heterogeneity in task composition within titles.
The central criticism is methodological: job-title-level assessment ignores the within-occupation task variation that substantially changes the risk picture.
7. According to David Autor's 2024 research, AI wage gains disproportionately flow to workers who:
Correct. Autor's insight: gains flow to workers who position as the rare human complement to AI — not to those avoiding or competing with it.
Incorrect. Autor found gains go to workers who amplify their rare capabilities through AI — the scarce human component in a human-AI system creating more value than either alone.
8. Denmark's flexicurity model is notable for combining:
Correct. Flexicurity pairs employer flexibility (relatively easy dismissal) with worker security (generous benefits — up to 90% wage replacement for two years) and active labor market programs (required participation in retraining/job search). It costs roughly 2% of Danish GDP annually.
Flexicurity is a "three-way" model: flexible hiring and firing rules, generous income security for unemployed workers, and active labor market programs that keep workers engaged in skill development. It explicitly does not offer lifetime employment or a UBI.
9. Goldman Sachs's GS AI platform data showed that the most valuable analysts were those combining what?
Correct. Goldman's data showed domain expertise + AI fluency produced 30–40% more client deliverables quarterly — the AI leverage metric.
Incorrect. Goldman found domain expertise plus AI fluency was the winning combination, producing measurably higher output per analyst.
10. What maximum fine can the EU AI Act impose on companies that violate its provisions for high-risk AI?
Correct. High-risk AI violations carry fines up to 3% of global annual turnover. Violations of outright prohibitions (e.g., social scoring) carry up to 7%.
For high-risk AI violations, the EU AI Act sets fines up to 3% of global annual turnover — significant enough to affect major technology companies' behavior globally.
11. David Deming's research found which category of workers showed the strongest wage and employment growth from 1980 to 2020?
Correct. The dual-skill combination — math plus social — showed the highest wage premium growth, reflecting complementarity between these skills and the demand for both in complex roles.
Deming's finding was specifically about combination — math plus social skills outperformed either in isolation, suggesting the synthesis is where the labor market premium concentrates.
12. In the Stockton SEED pilot, what percentage of monthly spending went to alcohol and tobacco?
Correct. Less than 1% of SEED spending went to alcohol or tobacco — one of the most commonly cited findings to counter the objection that cash transfers will be "wasted" on vices. Food, merchandise, and utilities accounted for the overwhelming majority.
Less than 1% of Stockton SEED spending went to alcohol or tobacco. This was a landmark finding because it directly refuted a common argument against unconditional cash transfers.
13. The "data labor" framework argues that workers displaced by AI have a claim to compensation because:
Correct. The data labor concept, associated with Jaron Lanier and elaborated in Posner & Weyl's Radical Markets, holds that since AI is trained on human-generated data, those humans performed a form of unpaid labor. Compensating it — through data dividends or micro-payments — would distribute AI productivity gains broadly.
Data labor theory holds that AI systems derive their value from training data generated by millions of ordinary people — who are currently uncompensated. Recognizing and paying for this contribution would create a mechanism for distributing AI gains, funded by the companies that benefit from the data.
14. After VisiCalc and spreadsheet software automated routine arithmetic in accounting (from 1979 onward), U.S. accounting employment:
Correct. The spreadsheet example is a classic complementarity case — automating routine arithmetic made financial analysis cheaper and more accessible, which expanded demand for the analytical and advisory tasks that accountants then spent more time on.
The spreadsheet case illustrates the complementarity effect: automating routine tasks expanded the market for analysis, increasing rather than decreasing employment.
15. Goldman Sachs' internal analysis estimated that generative AI could automate what proportion of work tasks at legal, accounting, and consulting firms?
Correct. 25–50% — a significant but not total displacement estimate, focused on specific task categories like document review, first-draft writing, and data summarization.
Goldman's estimate was 25–50% of work tasks — a substantial fraction but not total replacement, concentrated in routine knowledge tasks.
16. David Autor's "task recomposition" describes a process where AI:
Correct — task recomposition redistributes the task bundle toward what humans do distinctively well.
Task recomposition means AI absorbs routine cognitive sub-tasks within a job, concentrating the remaining work on judgment, creativity, and relationship-intensive activities.
17. Germany's co-determination (Mitbestimmung) system is relevant to AI displacement because it:
Correct. Mitbestimmung means workers participate in governance decisions at the firm level, including technology adoption. The Volkswagen AI transition included union negotiation two years before implementation — a direct result of co-determination requirements.
Co-determination gives workers formal governance roles — seats on supervisory boards and works council representation — with legally enforceable consultation rights on major decisions like AI adoption. This is structurally different from bans or quotas.
18. BT Group's 2023 announcement of 55,000 job cuts alongside 3,000 technology specialist hires reveals what structural reality?
Correct. The 10–18:1 displacement-to-creation ratio means the new roles are scarce and valuable — making deliberate positioning essential rather than optional.
Incorrect. BT's numbers reveal roughly 10–18 displacements per new role created — making the new specialist positions scarce and the positioning competition intense.
19. The New York attorneys who submitted AI-generated legal citations in 2023 faced sanctions because:
Correct. The professional responsibility to verify submissions before filing was the attorneys' — the source of the content (AI or otherwise) does not transfer that obligation elsewhere.
The attorneys faced sanctions because verification was their professional obligation — the AI generating the content did not transfer or reduce that duty.
20. Which of the following organizations used AI primarily as a "post-hoc reviewer" of human-drafted documents?
Correct — Allen & Overy deployed Harvey AI to review associate-drafted contracts after the human draft was complete, checking against pattern libraries and firm standards.
Allen & Overy's Harvey AI deployment used the post-hoc reviewer model — human associates draft contracts, AI then reviews against firm standards and flags unusual risk clauses. The human judgment produced the draft; AI checked it systematically.