David Shapiro critiques Anthropic’s collaboration with Wall Street and other powerful institutions, arguing that this “elite capture” concentrates AI development and benefits among a few dominant players, reinforcing existing power imbalances and limiting public access. Despite these concerns, he remains hopeful that with public investment, democratic governance, and antitrust measures, AI can be steered toward more equitable and inclusive outcomes.
In this video, David Shapiro discusses concerns about Anthropic, an AI company, developing financial models for Wall Street, which raises questions about the company’s commitment to AI safety. He suggests that Anthropic, influenced by effective altruists and rationalists, believes that collaborating with powerful institutions like governments, militaries, and Wall Street is morally justified if it helps prevent a catastrophic AI future, even if it means enabling a dystopian cyberpunk scenario. This perspective, according to Shapiro, reflects a form of “elite capture,” where AI development and benefits become concentrated in the hands of a few powerful entities rather than serving the broader public.
Shapiro outlines the characteristics of elite capture in AI, including control over critical infrastructure like chips, cloud data centers, and energy, which are dominated by a few major players such as Nvidia, TSMC, and leading cloud providers like AWS, Microsoft Azure, and Google Cloud. He highlights vertical integration, where companies like Anthropic control most aspects of AI development except chip manufacturing, and stratified access, where the best AI models are reserved for high-paying enterprise clients in finance and defense. This creates significant barriers for public and open-source alternatives, which lag behind in scale and capability.
The video also explores how AI is being deployed in ways that reinforce existing power imbalances. Examples include AI tools used by landlords to coordinate rent hikes illegally, workplace surveillance systems like Amazon’s time-off-task algorithm, and biased AI hiring tools that perpetuate discrimination. In finance, AI models provide elite investors with superior trading insights, further widening the gap between institutional and retail investors. Additionally, AI’s integration into national security through large defense contracts consolidates power among a narrow group of tech firms and government elites, limiting democratic oversight.
Shapiro emphasizes that the current AI landscape is shaped by massive capital investments from a handful of tech giants, who also influence regulatory frameworks to their advantage. Public AI infrastructure and governance are notably absent or underfunded, with public initiatives dwarfed by private spending. This concentration of power and resources means that AI’s benefits are not automatically distributed equitably but depend heavily on who controls the technology and the institutions around it. The pattern mirrors historical examples where technological advances initially benefit elites before broader diffusion occurs through regulation and public intervention.
Despite the grim analysis, Shapiro remains cautiously optimistic, believing that technology can ultimately raise living standards for all if countervailing institutions are built in time. He advocates for public compute resources, robust antitrust enforcement, democratic governance, and enhanced worker power to counterbalance elite capture. His work on post-labor economics focuses on rebalancing power in the AI era. The video concludes with a call to recognize the challenges ahead while maintaining hope that collective action can steer AI development toward more inclusive and equitable outcomes.