The Problem with AI Is the Problem with Capitalism (jacobin.com)

The piece argues that the immediate harms of powerful generative AI — from image models trained on artists’ work without consent to ChatGPT and GPT‑4 writing books, legal documents, code and convincing deepfakes — are not primarily technological but structural. The author catalogs concrete risks: rapid drops in the market value of creative and knowledge work, large‑scale job displacement across paralegals, programmers, customer service and more, faster production of realistic misinformation, and corporate incentives that prioritize profit over public good. Technical advances (better generative models, prompt engineering, large‑scale scraping for training data) make these harms practical and accelerating. Its core claim for the AI/ML community is that these problems stem from how capitalism allocates value and power, not from AI itself. Introduced under a different economic model automation could be liberating rather than destructive — e.g., shared gains, automation pensions or universal support rather than mass precarity. For practitioners and policymakers this implies focusing beyond model accuracy: who owns training data, deployment incentives, governance, redistribution mechanisms, and societal safeguards. The takeaway is that technical fixes alone won’t prevent exploitation; aligning AI development with public-interest economic and regulatory structures is essential.