Chapter 1: The Evolution of Artificial Intelligence

As a lawyer specializing in intellectual property, I have closely followed the advancements in artificial intelligence (AI) for many years. My fascination began four decades ago when I encountered an AI program designed for my Atari computer. Though it was rudimentary compared to today’s standards, it allowed users to enter conversational queries and receive basic responses.

Fast forward to 1997, when IBM's "Deep Blue" made headlines by defeating world chess champion Gary Kasparov. At that time, I could never have envisioned the extent to which AI would evolve—enabling machines to autonomously navigate celestial bodies, allowing military drones to coordinate attacks without human input, or even generating surreal images, such as a former president with six fingers.

Businesses are increasingly leveraging AI for tasks such as customer service, content creation, and marketing, all while reducing costs and increasing efficiency. Students utilize AI for academic research, language acquisition, and learning various subjects, while artists incorporate AI-generated visuals into their works. I too harness AI for hashtag suggestions and SEO-optimized social media posts, although I refrain from using it to produce content for my writing.

Despite the numerous benefits AI offers, there are significant drawbacks. A lawyer recently faced repercussions after relying on AI for legal research, only to discover that the case he cited was fictitious. Additionally, companies creating brand names with AI risk infringing on existing trademarks. In 2024, Air Canada's AI customer service system disseminated incorrect reimbursement information to a grieving customer, resulting in damages awarded against the company. Furthermore, Sports Illustrated faced backlash in 2023 for publishing articles attributed to fictitious authors, all generated by AI.

The list of AI mishaps continues to grow, but perhaps the most troubling aspect is its intersection with race. In a disturbing incident in Maryland, a disgruntled school employee employed AI voice manipulation to clone his principal’s voice, creating a fake audio clip that falsely portrayed the principal making racist and antisemitic remarks. This incident not only harmed the principal's reputation but also diverted attention from genuine issues of racism.

Research indicates that AI systems frequently struggle to accurately identify images and speech patterns of nonwhite individuals. In particular, facial recognition software shows a significant bias against Black individuals, leading to instances of mistaken identity within law enforcement—a serious concern given the existing biases in the criminal justice system.

AI programs like Stable Diffusion have been found to generate images that reflect racial and gender biases even more pronounced than those present in society. An analysis of over 5,000 images produced by this system revealed a stark underrepresentation of women and Black individuals in professions such as doctors, lawyers, and judges.

In one experiment, scientists tasked AI robots with identifying criminals from blocks featuring various faces. Alarmingly, the robots disproportionately identified a Black man's face as a criminal. Similarly, in the medical field, AI tools used for diagnosing skin conditions have shown significant racial bias, particularly in dermatology, as they have not been effectively programmed to address the needs of darker-skinned individuals.

The biases embedded in AI stem from the backgrounds and experiences of their creators. A glaring issue is that the majority of those developing AI technologies are white men. In 2023, a prominent AI conference saw over 90% of its 5,000 attendees being white men. This lack of diversity contributes to the perpetuation of societal biases within AI systems.

It is crucial for AI companies to confront and rectify these ingrained biases. As we strive to address societal inequities, it is imperative that these technologies are developed with inclusivity in mind. The injustices faced by Black individuals throughout history should not be compounded by technology that reflects the same prejudices.