Step by Step
D
Displacement — routine cognitive tasks automated
AI automates routine cognitive tasks like data entry, basic customer service, and simple analysis, displacing certain types of jobs even as it transforms and creates others.
Example: routine data-entry positions being significantly reduced as AI systems take over much of that repetitive work.
P
Power concentration — compute, data, and talent concentrated
Because AI development requires massive compute resources, enormous datasets, and specialized talent, this power tends to concentrate in a small number of large, well-resourced companies.
Example: only a handful of large tech companies having the resources to train truly cutting-edge, massive-scale AI models.
M
Misinformation — deepfakes and synthetic propaganda
AI enables the creation of deepfakes, synthetic media, and AI-generated propaganda at a scale and realism that was previously impractical.
Example: a convincing deepfake video of a public figure saying something they never actually said, generated using AI tools.
E
Environmental cost — massive energy consumption
Training large AI models consumes significant energy: GPT-3's training alone consumed approximately 500 tons of CO2 equivalent, comparable to roughly 60 transatlantic flights, with inference at scale adding up further over time.
Example: the total carbon footprint of training a single large frontier AI model being comparable to dozens of long-haul international flights.
Applied Walkthrough
1
A policy analyst is evaluating the broader societal implications of widespread AI adoption in a particular industry.
2
They consider labor displacement (which specific jobs are being automated versus transformed versus newly created), power concentration (which companies control the underlying AI capabilities), misinformation risks (how the technology could be misused for synthetic media), and environmental costs (the energy and carbon footprint of training and running these systems).
3
Each of these four dimensions represents a genuinely distinct societal impact category, requiring different policy responses and considerations.
4
The specific GPT-3 example — roughly 500 tons of CO2 equivalent, comparable to 60 transatlantic flights — illustrates just how significant the environmental dimension alone can be for even a single large model's training run.
Exam Application
Exams test whether you can name and distinguish these four broad societal impact categories (labor displacement, power concentration, misinformation, environmental cost) and whether you know the specific GPT-3 environmental impact figure (roughly 500 tons of CO2, equivalent to 60 transatlantic flights) as a concrete, frequently cited example.
⚠ Common Trap
The most common trap is treating AI's labor impact as simple, one-directional job destruction. The framing in this lesson is more nuanced: AI transforms labor, with some jobs lost, some changed, and new ones created — not simply pure job elimination.
✓ Quick Self-Check
1. What kind of jobs does AI-driven labor displacement primarily affect?
Routine cognitive tasks like data entry, basic customer service, and simple analysis.
Tap to reveal / hide
2. Why does AI development tend to concentrate power in a small number of companies?
Because it requires massive compute, enormous datasets, and specialized talent, resources concentrated among few large, well-resourced organizations.
Tap to reveal / hide
3. What kind of misinformation risk does AI specifically enable at scale?
Deepfakes, synthetic media, and AI-generated propaganda.
Tap to reveal / hide
4. Approximately how much CO2 did training GPT-3 consume, and what is that comparable to?
Approximately 500 tons of CO2 equivalent, comparable to about 60 transatlantic flights.
Tap to reveal / hide
5. Does AI's impact on labor mean pure job elimination, according to this lesson?
No — some jobs are lost, some are changed, and new ones are created; AI transforms labor rather than simply replacing it.
Tap to reveal / hide