Amazon has come under fire once again for faulty AI systems, this time for racial and gender bias. Since November of this past year, Amazon has been defending its AI system after reports claimed that the software was being sold to law enforcement for random surveillance in public spaces. Eight members of Congress wrote CEO Jeff Bezos a letter expressing their concern, highlighting that the product “places disproportionate burdens on communities of color, and could stifle Americans’ willingness to exercise their First Amendment rights in public.” In addition, the ACLU wrote Bezos a letter of disapproval, Google withdrew from government sales of the AI system, and Microsoft called for stricter regulations on the product.
Still, Amazon has refuted all claims, with GM for Deep Learning and AI Dr. Matt Wood defending the system and accusing others of not understanding the entirety of the technology. According to the researchers, Amazon's AI was accurate in inferring the gender of light-skinned males and inaccurate in inferring the gender of darker-skinned females. Dr. Wood responded, “Across all ethnicities, we found no significant difference in accuracy with respect to gender classification. In a broader test of facial recognition, we...found exactly zero false positive matches at the recommended 99% confidence threshold.” AI in itself is controversial, and Amazon has vowed to keep improving the product until there are no doubts.
