Free tools Windows power users keep installed
One-click scans. No signup required.
iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Capitalism can shape which AI systems get built, who controls them, and who gains or loses from their use. But the claim that AI is only bad because of capitalism goes further than the available evidence: AI outcomes also depend on human decisions, system design, governance, and purpose. The more defensible question is how economic incentives and institutional power steer AI—and what changes could make its uses more accountable and broadly beneficial.
What does it mean to say AI is “bad”?
“Bad” can refer to very different outcomes: intrusive surveillance, unfair or unchallengeable decisions, pressure on workers, benefits concentrated among owners, or environmental costs. These outcomes should not be collapsed into one claim about AI as a whole. A system’s effects depend on what it is designed to do, the data and assumptions behind it, who deploys it, and what rules govern its use.
AI is not an independent actor choosing its own social role. In The Political Philosophy of AI: An Introduction (2022), Mark Coeckelbergh emphasizes that AI’s political effects are mediated by people and institutions: people select data, set goals, decide where to apply systems, and determine how to respond to their outputs. That does not make the technology irrelevant; it means responsibility cannot be assigned to the technology alone.
How can capitalism shape AI’s effects?
The strongest political-economic argument is about incentives and control, not a claim that every AI system produces harm. If infrastructure and data are controlled by a small set of organizations, and investment is driven by expected returns, those organizations can have considerable influence over which problems AI is built to solve and how its gains are distributed. Critical accounts connect these arrangements to several possible harms.
Profit can influence what gets built
Organizations are more likely to fund uses they expect to benefit from. That can direct development toward commercial priorities even when other applications might better serve workers or communities. The European Trade Union Institute’s 2024 volume Artificial Intelligence, Labour and Society describes a policy frame focused on economic competitiveness alongside a frame focused on societal challenges. The distinction matters: a technology’s value depends partly on whose objectives guide its development and deployment.
Ownership and workplace power can shape labor outcomes
When employers control AI tools and the data they generate, systems can support worker tasks, but they can also expand monitoring or give management more control over the pace and evaluation of work. Whether productivity gains improve workers’ conditions, reduce staffing, or accrue mainly to owners is an institutional and political question—not an automatic consequence of using AI. The political-economy critique identifies labor substitution and control as mechanisms to examine, not universal outcomes already established for every workplace.
Unequal control can make decisions hard to challenge
If a person affected by an AI-assisted decision cannot learn how it was made, correct relevant information, or obtain meaningful review, the problem is not just technical accuracy. It also concerns who has authority, who can demand an explanation, and who can provide a remedy. Concentrated control over systems and data can make those power imbalances harder to contest.
Business incentives can contribute to environmental costs
AI is not immaterial software. The ETUI volume’s environmental chapter calls for examining impacts across the lifecycle and supply chain, including energy use and emissions, material extraction and toxicity, water demand for data-center cooling, and electronic waste. The chapter also analyzes how consumption and business incentives can affect demand. These are important categories for assessment, but they do not add up to one universal impact figure for AI; effects vary with the system, its scale, its energy and materials, and how equipment is handled at the end of its life.
Why “only because of capitalism” is too strong
The word “only” makes a counterfactual claim: that the harms attributed to AI would disappear if the economic system changed. The sources discussed here do not establish that. They support analyzing how ownership, profit incentives, labor relations, and institutional power influence AI outcomes, but they do not provide a controlled comparison showing what would happen under a non-capitalist system.
Changing economic arrangements could alter who owns computing infrastructure, who chooses system goals, and how benefits and costs are shared. It would not by itself guarantee sound design, accurate decisions, privacy, accountability, or protection against coercive uses of technology. Those questions also depend on governance, technical choices, and the ability of affected people to challenge decisions.
Rank #3
Nor does the existence of commercial incentives prove that all AI uses are harmful. The ETUI volume presents AI as a technology that can be directed toward economic competitiveness or societal challenges. Coeckelbergh’s account also leaves room for public-interest uses while emphasizing human accountability and political judgment. Purpose and context matter; neither “AI is inherently harmful” nor “AI is automatically beneficial” follows from the sources.
What would a fairer AI system need to change?
Instead of treating “capitalism” as a complete explanation, assess the institutions around a particular system. These questions make the political-economic argument concrete and help distinguish a problem with a specific deployment from a broader problem of ownership or incentives.
- Ownership and control: Who owns or controls the infrastructure and data, and who can set the system’s goals?
- Purpose: What task is the system intended to perform, and whose interests does that purpose serve?
- Participation: Can workers and affected communities influence design and deployment decisions?
- Distribution: Who receives productivity gains, and who bears the risks, costs, or disruption?
- Accountability: Can people understand, contest, and seek a remedy for decisions that affect them?
- Environmental impact: Are impacts measured across production, operation, and disposal, rather than only while software is running?
These tests do not prescribe one economic system. They identify the decisions and power relationships that any proposed reform would need to address. A system might be privately owned, publicly operated, or governed through another arrangement; the relevant question is whether its purposes, effects, and decisions are open to meaningful scrutiny and accountability.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What the evidence can—and cannot—say
The scholarship cited here offers frameworks for analyzing power, labor, policy goals, and environmental costs; it does not settle every empirical question about AI. For example, the ETUI chapter discusses lifecycle categories rather than establishing one figure that represents all AI’s environmental impact. It also mentions an estimate that the AI-in-oil-and-gas market would reach around USD 4 billion globally by 2025, compared with USD 1.75 billion in 2018. That was a market forecast, not a verified 2025 result or a measure of AI’s overall social or environmental effects, so it cannot support a general claim about AI’s costs or benefits.
A 2019 response published by Elon University’s Imagining the Internet Center also illustrates why AI’s future is often discussed in relation to broader economic and political conditions. An anonymous professor emeritus at a U.S. university school of information wrote, “I’m unable to reflect on these questions about the future of AI separate from the context and trends in global economics, governance and politics.” This is one expert’s response, not a consensus finding or a measured prediction. It is a useful prompt for analysis, not proof that any single economic system determines AI’s effects.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteA more defensible version of the claim
“AI is only bad because of capitalism” is best treated as a provocative thesis, not a conclusion established by the evidence. Its strongest defensible core is that economic incentives and ownership can steer AI toward surveillance, workplace control, and unequal distribution of gains and costs. Whether those harms occur—and whether reforms reduce them—depends on institutional choices, technical design, deployment purpose, and governance as well.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
