Descriptions:
UCLA professor Safiya Noble, author of the widely cited book “Algorithms of Oppression,” joins Bloomberg Technology to argue that the AI industry is moving too fast and that current large language models are fundamentally unready for the institutional deployments being pursued at scale.
Noble’s critique centers on three compounding problems. First, LLMs are unreliable: factual errors are pervasive enough that companies must employ humans to verify outputs, eroding the labor-cost savings that justified adoption in the first place, and she contends corporate America is already quietly pulling back from AI chatbot investments. Second, the training data embeds the full spectrum of societal discrimination — racial bias, gender bias, geographic and political skew — and the models obfuscate this inequality by presenting outputs with false confidence and apparent authority. Third, the engineers building these systems lack the sociological and historical literacy to recognize or correct for what their data encodes.
Drawing on her earlier research into commercial search engines, Noble argues that LLMs repeat the same structural failure at greater scale and with higher stakes, particularly as public institutions, schools, and libraries consider making them foundational infrastructure. On solutions, she advocates redirecting investment toward smaller, purpose-built models developed with input from underrepresented communities, and preserving human expertise in journalism, education, and research rather than substituting machine-generated content.
📺 Source: Bloomberg Technology · Published June 10, 2026
🏷️ Format: Interview
