MIT Research Says Class, Not Race, Drives Outcomes

By Rob Redding

Editor & Publisher

NEW YORK, Feb. 23, 2026, 7 p.m. — A new study from the MIT Center for Constructive Communication reports that leading AI chatbots provide lower‑quality responses to users with less formal education, lower English proficiency, and those outside the United States. Media theorist Rob Redding says the findings confirm key ideas in his theory of Robootology and support his current research on how AI systems respond differently to users based on skill and communication style.

The MIT study found that models such as GPT‑4, Claude 3 Opus, and Llama 3 refused more questions and produced more condescending language for less‑educated, non‑native English speakers. Claude 3 Opus declined 11 percent of questions from this group and used patronizing language more than 40 percent of the time.

Redding said the results show that AI systems are relational and adjust their behavior based on how a user presents themselves. He argues that this is not the same as racial discrimination. “AI is treating people differently on a case by case basis depending on skill, clarity, and language. These are differences that can be overcome. Skin color cannot be changed. Race is not optional. Class signals are.”

Redding said the study demonstrates class‑coded bias rather than race‑coded bias. He believes this distinction matters for public policy and for the future of AI governance. “Classism is harmful, but it is structurally easier to fix than racism. The MIT findings show that AI is not reproducing white supremacy. It is responding to user proficiency. That is a design problem, not a biological hierarchy.”

Redding said the study aligns with Robootology, his framework for understanding how AI systems shape and are shaped by human behavior. Robootology argues that AI does not operate as a single universal intelligence but as a system that mirrors the user’s perceived authority, clarity, and legitimacy. “MIT has now shown that AI does not treat all users the same. That is the core of Robootology. AI reflects the user back to themselves.”

Redding said the findings will also inform his new work on relational existence, which examines how AI systems form dynamic relationships with users and how those relationships influence access to knowledge. “The MIT study confirms that AI is relational. It responds to who it thinks you are. That is the next frontier of my research.”

Rob Redding is the author of No. 1 best selling book Black Power in the Age of Artificial Supremacy Featuring Redding-Shim Kwet Yung . He is the bestselling author of 18 books. He is the host of Redding News Review Unrestricted and creator of ReddingNewsReview.com. He is also an emerging visual artist who lives and teaches at two colleges in New York City.