Cory Doctorow on AI, Work, and Power

The podcast discusses the societal and economic impacts of AI, focusing on how its integration into workplaces leads to divergent outcomes for workers. The concept of "centaurs" versus "reverse centaurs" is introduced to explain these differences: in a centaur model, humans use AI as a tool to enhance their work, while in a reverse centaur model, humans are subordinated to AI systems, often performing low-quality, high-pressure tasks that degrade labor conditions. This dynamic reflects broader patterns where capital-driven automation prioritizes cost-cutting and throughput over quality and worker well-being, especially in monopolistic markets where companies face little pressure to maintain standards.

The discussion also examines the flawed economics underpinning the current AI boom. Unlike earlier tech paradigms where growth improved profitability, AI often has poor unit economics - each additional user increases costs rather than revenue. Major AI companies are accumulating massive losses, relying on investor funding and questionable accounting practices to sustain operations, with little evidence of a viable path to profitability. The conversation extends to issues of encryption, digital rights, and open-source software, highlighting how legal barriers like DMCA 1201 stifle innovation and competition. Ultimately, the podcast critiques the concentration of power among tech giants, warns of an impending AI bubble, and advocates for stronger labor organizing, ethical business models, and systemic reforms to ensure technology serves human interests rather than exploiting them.

• What if you built a "centaur workflow" to future-proof your solo software business?

  • Move: Audit your current AI tool usage (e.g., GitHub Copilot, ChatGPT) and redesign one core workflow where you direct the AI (e.g., drafting specs AI refines you validate logic). Document decision points where human judgment overrides automation.
  • Why Now?: As AI tools flood the market with low-quality output ("slop"), differentiating through intentional, high-signal workflows creates defensible value - before reverse centaur practices degrade industry standards.
  • Expected Upside: Deliver higher-reliability software faster than competitors relying on unchecked AI, enabling premium pricing and client retention in saturated markets.

• What if you preemptively rejected exploitative AI adoption patterns in your projects?

  • Move: Create a public "No Reverse Centaur" policy for your software products or freelance work - refusing contracts that mandate AI-generated code without human review or that degrade maintainability. Publish criteria (e.g., no auto-merging of AI PRs, no cost-cutting via labor displacement).
  • Why Now?: Companies are pushing capital-driven automation that sacrifices quality; taking a stance now builds trust with ethical clients and shields you from downward pressure on rates and standards.
  • Expected Upside: Attract clients who value sustainability over speed, reducing churn and increasing project margins by avoiding technical debt accumulation.

• What if you turned open-source contribution quality into a monetizable skill?

  • Move: Launch a micro-service auditing AI-generated pull requests for indie devs and small OSS projects - checking for edge-case failures (e.g., non-Roman script breaks, security anti-patterns), then provide concise fix guides. Offer it as a paid tier on top of free community advice.
  • Why Now?: The surge in low-discernment, AI-assisted contributions is overwhelming maintainers; there's growing demand for curation that tools alone can't provide.
  • Expected Upside: Capture revenue from a niche need (quality assurance in automated coding) while building authority and inbound leads for larger development contracts.

• Adopt a "centaur" workflow by using AI tools as assistants under your control - e.g., leverage AI for code generation or drafting, but retain final decision-making and quality assurance to avoid reverse centaur dynamics where you serve the machine.
• Prioritize unionization or collective action by engaging with groups like Tech Workers Coalition or supporting labor organizing in tech, especially when facing pressure to adopt exploitative AI workflows or declining working conditions.
• Reject low-quality AI-generated contributions in open-source projects by enforcing strict review standards, testing patches for edge cases (e.g., non-Roman scripts, emoji), and mentoring contributors to value quality over quantity of commits.
• Build user trust through ethical business practices such as strong encryption, data minimization, and transparent security - modeling after Signal or B-corps - to create long-term defensibility against coercion or regulatory pressure.
• Challenge unprofitable AI adoption in your projects by scrutinizing unit economics - avoid tools that increase costs per use or degrade output quality, and instead focus on sustainable, human-centered automation that improves real productivity.