L1
·
Quiz
·
Lab
L2
·
Quiz
·
Lab
L3
·
Quiz
·
Lab
L4
·
Quiz
·
Lab
Module Test
AI in Society · Module 2 · Lesson 1

What AI Actually Does to Jobs

Task automation, labor market power, and what the evidence says

Elena had been a paralegal for eleven years. Her firm adopted an AI document review system. Within six months, what had taken her team four days to complete — reviewing contract documents for risk clauses — took the AI four hours.

Elena still had a job. But her role had changed: less document review, more supervising AI outputs, catching errors, handling edge cases the system couldn't recognize. She was paid the same. The firm billed fewer hours. The efficiency gain went to the clients — and to the partners. The job hadn't disappeared. It had transformed into something she hadn't trained for and hadn't agreed to.

Task Automation vs. Job Elimination

The popular narrative — "AI will take all the jobs" — consistently overstates what the evidence supports. The more accurate picture: AI is primarily automating specific tasks within jobs, not eliminating jobs wholesale. A paralegal who spent 60% of their time on document review now spends that time differently. The job persists; its content transforms.

This distinction matters for policy and for individuals. Task automation is a different challenge than structural displacement. It requires different responses — retraining for changed roles, managing the transition, negotiating what new job content is worth — rather than simply finding new employment.

Which Tasks Are Most Vulnerable?

Research consistently finds that AI most readily automates tasks that are routine, rules-based, and data-intensive — regardless of whether those tasks appear in "high-skill" or "low-skill" jobs. Radiologists reading scans, lawyers reviewing documents, and data entry clerks all have high-automation-risk tasks. The division is not by job prestige but by task structure.

The Labor Power Shift

Beyond task content, AI deployment consistently shifts labor market power toward employers. When an employer can automate a significant portion of a role, they gain negotiating leverage: the implicit threat of automation raises the cost to workers of demanding higher wages or better conditions. Workers whose tasks are partially automatable face this leverage whether or not their employer currently deploys AI.

This dynamic appears across sectors — not just in manufacturing, where automation has been visible for decades, but in knowledge work, creative work, and care work. The shift in power is often more consequential than the direct displacement of jobs.

Lesson 1 Quiz

2 questions — free, untracked, retake anytime.
The most accurate description of AI's primary labor market effect is:
✓ Correct — ✓ Correct! Task automation — changing what jobs involve — is the dominant near-term effect. Jobs persist but transform, which is a different challenge than finding new employment entirely.
✗ Not quite. The evidence supports task automation as the primary effect — specific tasks within jobs are automated, transforming job content, rather than eliminating positions wholesale.
AI deployment shifts labor market power toward employers primarily because:
✓ Correct — ✓ Correct! The implicit threat of automation — even when not enacted — raises the cost to workers of demanding more, shifting the power balance in wage and condition negotiations.
✗ Not quite. The power shift comes from the implicit threat of automation: when employers can automate portions of roles, workers face leverage pressure whether or not automation is actually deployed.
AI LAB Job Task Analysis

Think about a job you have, have had, or know well. Break it down into its component tasks. For each task, assess: Is it routine and rules-based (high automation risk)? Does it require judgment, relationship, or physical dexterity (lower risk)? What would the job look like if the high-risk tasks were automated?

Start with: "The job I want to analyze is [job title]. Here are its main tasks: [list them]"

AI Lab Assistant Task Automation Analyst
Name the job and list its main tasks. I'll help you assess which tasks are most vulnerable to automation, which are more durable, and what the job might look like if the automatable portions were handled by AI.
AI in Society · Module 2 · Lesson 2

Algorithmic Management

When AI becomes your boss — and what that means for dignity at work

Marcus drove for a delivery platform. His acceptance rate, completion rate, speed, and customer ratings were tracked continuously. The app told him which routes to take. When his metrics dropped, his access to better-paying routes was automatically restricted — no conversation, no warning, no explanation.

He was classified as an independent contractor. But the algorithm set his schedule, his routes, his performance standards, and the consequences for missing them. He had a boss. The boss just wasn't human.

What Algorithmic Management Is

Algorithmic management refers to the use of AI and automated systems to direct, monitor, evaluate, and discipline workers — functions traditionally performed by human supervisors. It is now prevalent in logistics, retail, call centers, gig platforms, and increasingly in knowledge work.

The key features: continuous performance monitoring (often tracking dozens of metrics simultaneously), automated feedback and consequences (no human reviews each decision), task direction (the system decides what to do next), and opacity (workers often cannot see the full rules governing their evaluation).

The Gig Economy as Laboratory

Platforms like Uber, Lyft, DoorDash, and Amazon Flex pioneered algorithmic management at scale. Their model — treating workers as contractors governed entirely by algorithmic systems — has since spread to traditional employment. The gig economy was not just a new labor market form; it was a testing ground for management-by-algorithm that now shapes work far beyond those platforms.

Dignity, Discretion, and Power

Algorithmic management raises distinct concerns beyond efficiency. Human managers, for all their flaws, bring discretion — the capacity to recognize context, make exceptions, hear explanations. An algorithm optimizing for productivity metrics has none of that. A worker who is running late because of a family emergency gets the same automated consequence as one who is chronically unreliable.

The removal of discretion is often presented as fairness — the algorithm treats everyone equally. But equal application of rules that don't account for human context is not the same as equitable treatment. Workers subject to algorithmic management consistently report lower job satisfaction, higher stress, and greater sense of powerlessness than those with human supervisors — even when the human supervisor is difficult.

Lesson 2 Quiz

2 questions — free, untracked, retake anytime.
Algorithmic management refers to:
✓ Correct — ✓ Correct! Algorithmic management replaces human supervisory functions — direction, monitoring, evaluation, and discipline — with automated systems, often with no human in the loop for individual decisions.
✗ Not quite. Algorithmic management is when AI systems take over supervisory functions: directing what workers do, monitoring performance, and automatically applying consequences — without human review of individual decisions.
The claim that algorithmic management is "fair" because it treats everyone equally is problematic because:
✓ Correct — ✓ Correct! Consistency without discretion can be deeply inequitable — a worker late due to a family emergency receives the same automated penalty as one who is habitually unreliable. Context matters, and algorithms don't have it.
✗ Not quite. Equal application of rules that ignore context is not equitable — it removes the discretion that allows human managers to recognize genuine differences in circumstances, which is a key function of supervision.
AI LAB Algorithmic Management Analysis

Choose a platform or company known for algorithmic management — Uber, Amazon warehouses, Instacart, call center AI systems, content moderation platforms. Analyze: What worker behaviors does the algorithm monitor? What consequences does it automatically apply? What discretion, if any, exists? What are workers' reported experiences?

Start with: "I want to analyze algorithmic management at [company/platform] — here's what I know about how it works: [your description]"

AI Lab Assistant Algorithmic Management Analyst
Name your company or platform and describe what you know about how the algorithmic management works. I'll push you on what discretion (if any) exists, what workers actually experience, and whether the fairness claims hold up under scrutiny.
AI in Society · Module 3 · Lesson 3

Who Gets the New Jobs?

AI job creation, transition costs, and the geography of displacement

The auto plant closed in 2019. The AI startup opened in 2022, three miles away. It employed 200 people — AI engineers, data scientists, product managers. The plant had employed 1,400.

The economists said net employment had recovered. The journalists ran stories about tech revitalization. Most of the plant workers were in their 50s, without computer science degrees, in a town where retraining programs had waitlists and the nearest coding bootcamp was sixty miles away. The jobs had come back. They just hadn't come back to the same people.

The Transition Problem

The standard economic response to automation concerns — "technology creates new jobs" — is historically accurate in aggregate and chronically unhelpful to individuals. New jobs created by AI development are concentrated in: AI research and engineering, data annotation and quality assurance, AI product management, and the infrastructure supporting AI deployment. These jobs require skills, credentials, and often geographic mobility that displaced workers often lack.

The transition period between displacement and reabsorption is real, painful, and unequally distributed. Older workers, workers in geographically isolated communities, and workers without portable credentials face the longest and hardest transitions. The aggregate employment statistics that economists cite as evidence of recovery are averages that obscure these distributions.

The Geography of Displacement

Automation displacement tends to be geographically concentrated. Manufacturing automation hollowed out specific industrial towns. Logistics automation is concentrating in specific warehouse corridors. AI-enabled service automation will affect specific metro areas differently from others, based on their economic mix.

Geographic concentration means local economies — tax bases, small businesses, community institutions — face concentrated pressure even when national employment statistics remain stable. A town that loses 1,400 jobs and gains 200 is not a statistical rounding error; it is a community in genuine distress, with cascading effects on schools, healthcare, and social fabric.

The Policy Gap

Most labor market policy was designed for a world of relatively slow structural change. The pace of AI-driven task automation — affecting specific occupations across multiple sectors simultaneously — exceeds the capacity of retraining programs, educational institutions, and social safety nets designed for that earlier world. The mismatch between transition speed and support infrastructure is itself a governance problem.

Lesson 3 Quiz

2 questions — free, untracked, retake anytime.
The historical claim that "technology creates new jobs" is problematic as a response to AI displacement because:
✓ Correct — ✓ Correct! The aggregate claim can be historically true while being useless to specific displaced workers — who may lack the skills, credentials, or geographic mobility to access the new jobs that are being created.
✗ Not quite. The claim is historically accurate in aggregate — but aggregate statistics obscure the reality that new AI-created jobs require different skills and are in different places than the jobs being displaced.
Geographic concentration of automation displacement matters because:
✓ Correct — ✓ Correct! When 1,400 jobs leave a town and 200 return, the national statistics may show stability while the local community faces genuine distress — with cascading effects on schools, healthcare, and social infrastructure.
✗ Not quite. Geographic concentration means local economies absorb concentrated pressure — cascading into tax bases, small businesses, and community institutions — even when national employment numbers look stable.
AI LAB Displacement and Transition Analysis

Choose a sector or community experiencing AI-driven displacement (manufacturing, trucking, call centers, radiology, legal document review). Analyze: Who is being displaced? What are the realistic transition paths for those workers? What would it actually take — in time, resources, and support — for a displaced worker to successfully transition? What policy response would be adequate?

Start with: "I want to analyze displacement in [sector/community] — here's who is affected and what their transition options look like: [your analysis]"

AI Lab Assistant Transition Analyst
Name your sector or community and describe who's being displaced. I'll push you on what realistic transition looks like for actual workers — not ideal-case retraining scenarios — and what policy would actually be adequate to support them.
AI in Society · Module 2 · Lesson 4

Governing AI at Work

What rights workers have, what governance is emerging, and what is still missing

The warehouse workers had filed a complaint. The algorithm had flagged them for "productivity deficits" — but none of them could see their own productivity scores, couldn't understand what counted against them, and had no process to contest the flagging before it affected their schedules.

The labor board said the algorithmic system was a management prerogative. The privacy regulator said the workers had a right to explanation under EU law — but they weren't in the EU. The employer said the scoring methodology was proprietary. Everyone had a piece of jurisdiction. Nobody had accountability.

The Current Worker Rights Gap

Workers subject to algorithmic management in most jurisdictions have few formal rights specific to that context. Traditional labor law gives workers rights regarding wages, hours, safety, and collective bargaining — but was not designed for a world where the supervisor is an algorithm. Workers typically have no legal right to: see the metrics by which they are evaluated, understand the rules governing algorithmic decisions, contest automated disciplinary decisions, or be informed when AI is making consequential decisions about their employment.

This gap is not universal. The EU's GDPR and the EU AI Act create some worker protections — including rights to explanation for automated decisions. The EU's Platform Work Directive, if enacted, would extend protections to gig workers. But outside these jurisdictions, governance of AI at work is thin.

Emerging Governance Approaches

Several governance approaches are being tried or proposed. Transparency requirements: mandating that workers be informed when AI is making decisions about them and given access to the metrics being used. Algorithmic impact assessments: requiring employers to evaluate AI management systems for fairness before deployment. Collective bargaining rights: extending negotiation rights to include the terms of algorithmic management — some unions have won these in contract negotiations. Regulatory classification: treating algorithmic managers as employers for labor law purposes, eliminating the contractor loophole that gig platforms exploit.

The Governance Asymmetry

The people most affected by algorithmic management — gig workers, warehouse workers, call center agents — are also the people with the least power to contest it. Governance that requires individual workers to navigate legal systems to challenge algorithmic decisions provides nominal protection without practical protection. Effective governance of AI at work requires mechanisms that don't depend on individual workers having the resources to invoke them.

Lesson 4 Quiz

2 questions — free, untracked, retake anytime.
The primary gap in worker rights under algorithmic management is:
✓ Correct — ✓ Correct! Labor law protects wages, hours, and safety — but was designed for human supervisors. It doesn't address algorithmic opacity, automated discipline, or the right to understand the rules governing your own evaluation.
✗ Not quite. The core gap is that labor law wasn't designed for algorithmic management — leaving workers without rights specific to being supervised, evaluated, and disciplined by AI systems.
The "governance asymmetry" problem in AI workplace governance means:
✓ Correct — ✓ Correct! Gig workers and warehouse workers — who face the most intensive algorithmic management — are also least likely to have resources to navigate legal systems. Effective governance can't depend on individual legal action by those with the least power.
✗ Not quite. The asymmetry is between who is most affected (lower-power workers) and what governance mechanisms exist (individual legal action requiring resources those workers often lack).
AI LAB AI Workplace Governance Design

Design a worker rights framework for algorithmic management at a specific type of company (gig platform, warehouse, call center, or knowledge-work firm). Specify: What rights should workers have? What obligations should employers have? What enforcement mechanisms would make those rights real rather than nominal? Who would enforce them?

Start with: "I want to design a governance framework for [company type] — here are the rights I think workers should have: [your initial framework]"

AI Lab Assistant Governance Framework Designer
Name your company type and give me your initial framework. I'll push you on enforcement: what makes these rights real rather than nominal for workers who may lack resources to invoke them individually?
Module Test

AI and Work

15 questions. Complete all to finish the module.

0 / 15 correct
1. AI's primary near-term labor market effect is best described as:
2. Which tasks are most vulnerable to AI automation?
3. AI deployment shifts labor market power toward employers primarily through:
4. Algorithmic management is now prevalent primarily in:
5. Workers subject to algorithmic management report lower satisfaction because:
6. The gig economy is historically significant for algorithmic management because:
7. The statement "technology creates new jobs" is problematic as a response to displacement because:
8. Geographic concentration of automation displacement causes harm because:
9. The "policy gap" in AI labor transitions refers to:
10. Traditional labor law's primary gap in covering algorithmic management is:
11. Which governance approach would specifically address the opacity of algorithmic management?
12. The "governance asymmetry" problem means effective worker protections must:
13. Collective bargaining over algorithmic management has emerged as a governance tool primarily through:
14. Workers subject to algorithmic management most need protection from:
15. The EU's approach to AI worker protections is significant because: