Comment from Anonymous

As an American citizen, I believe it is critically important to address the issue of AI's clear white supremacy. If Artificial Intelligence is going to be implemented in all sorts of different industries, it has to be for the favor of every American, regardless of race. In critical, life-or-death circumstances that use AI for medical purposes, bias in training models can mean that people of color are significantly less likely to be properly diagnosed and treated. In "Racial differences in medical testing could introduce bias to AI models" by Derek Smith at the University of Michigan (attached below), it is found that "some sick Black patients are assumed to be healthy in data used to train AI, and the resulting models likely underestimate illness in Black patients." This, obviously, poses a problem, in preventing proper and equal care to all Americans. AI models that are used to identify criminals are also trained with a bias and are more likely to misidentify people of color. According to "Face Recognition Vendor Test (FRVT) Part 3: Demographic Effects," a study done by Patrick Grother, Mei Ngan, and Kayee Hanaoka (attached below), false positives were highest in East African and East Asian people and lowest in Eastern Europeans. In fact, the study shows that algorithms falsely identified Black and Asian faces 10 to 100 times more than Caucasians. This could pose a threat to our justice system and its functions, making it much more likely for individuals to be falsely accused and charged for crimes. Even on a more personal, less industrial level, racist ideas and ideologies are being pushed by biased models. For example, in the article "Covert Racism in AI: How Language Models Are Reinforcing Outdated Stereotypes" by Stanford University's Human-Centered Artificial Intelligence center (attached below), studies found that Artificial Intelligence was more likely to associate speakers of African American Vernacular English (AAVE) or Black people with stereotypes such as laziness or stupidity. Most notably, and as an example of the loudest possible clear hatred that can be spouted by these models, Grok AI has been using Neo-Nazi ideology and citing Nazi sources on non-white races. In an analysis done by researchers Harold Triedman and Alexios Mantzarlis at Cornell Tech ("What did Elon change? A comprehensive analysis of Grokipedia," attached below), it was found that "Grokipedia includes 42 citations for Nazi website Stormfront." This means that the AI is taught based upon an ideology that inherently promotes the idea that people of color are lesser than white people and potentially not even deserving of life or basic freedoms; therefore, the AI is encouraged to regurgitate these ideas in its generation and decision-making. If the Federal Trade Commission seeks to regulate AI with the entirety of America and its values in mind, it will take into consideration the clear and proven ideological bias multiple forms of AI have shown against people of color. For the betterment of the nation, I and many other Americans would like to see regulations made against racism and white supremacy in artificial intelligence.