Critical Race Theory gives us a set of analytical tools for understanding AI that technical and policy frameworks consistently miss:

Racism is structural, not individual. Algorithmic bias is not primarily a matter of bad intentions. It reflects the racial inequalities embedded in the data, systems, and social structures that AI learns from.

Race is a social construct. AI systems that treat racial categories as natural encode a false premise, and the discriminatory histories embedded in those categories.

Algorithmic accuracy and racial fairness are not the same thing. A system can be technically accurate and racially unjust simultaneously. The choice between them is political, not technical.

Surveillance and policing AI amplify structural racism through feedback loops that convert historical discrimination into apparently objective predictions.

Interest convergence explains when AI ethics reform happens, and whose interests it primarily serves. Reform that does not serve the interests of the most harmed communities will advance slowly, if at all.

Counter-storytelling and community voice are not supplementary to AI governance. They are analytically essential.

Racially just AI requires structural change, not just technical adjustment, attending to the social conditions that produce discriminatory data, and redistributing the power to design, oversee, and contest AI systems.

One mind cannot see it all. The minds that AI systems most consistently fail to see are not accidental omissions. Addressing them requires more than better algorithms. It requires confronting the social structures those algorithms reflect.