Amongst the past 25 years of engaging in AI projects, Lawrence, a Black man, lamented that his colleagues have seldom looked like him. He emphasized the need to challenge AI's promise to rapidly transform society by combating ingrained prejudices that may become encoded within new technologies.
Lawrence, the author of "Hidden in White View," which explored the part AI plays in systemic racism, criticized AI's reliance on data that could be inherently biased, riddled with racial, sexual, and faulty information.
In August, a Detroit mother was wrongfully arrested during the eighth month of her pregnancy. Surveillance technology had been used in an attempt to connect her to a crime. The Detroit Police Chief later attributed the situation to inadequate investigation.
A 2022 study revealed that AI-educated bots had a tendency to pair Black men with criminals or housewives with women. The research team concluded that continued use of these technologies may strengthen detrimental stereotypes and contribute to racism and sexism.
New York City’s health authorities recently formed a coalition to scrutinize clinical algorithms that adjust based on race, as they claim these algorithms often yield harmful consequences for people of color. According to an NYC Department of Health and Mental Hygiene statement, these algorithms overestimate the health condition of people of color, potentially leading to treatment delays.
OpenAI, the company behind ChatGPT and other AGI models, acknowledged in a statement shared with CNN that unconscious bias represented a significant challenge in the industry and that their commitment was to explore and reduce bias and other risks within their models.
Lawrence, believing that AI should reflect experiences of people of color, suggested involving them throughout the AI development process. He acknowledged that there were few Blacks or data scientists involved in the provision and shaping of AI solutions and that the only way to give them a seat at the table was to educate them.
Research has found that diversity and representation in the technology sector begins before college, as Code.org's Advocacy Coalition Report from 2023 revealed that Black and Hispanic students often have limited access to basic computer courses at high school. While 89% of Asian and 82% of White students had the opportunity, only 78% of Black and Hispanic students, and 67% of Indian students did.
Keeping diversity in AI at the forefront could make the technology safer and more ethical, according to Lawrence. He founded the nonprofit organization AI 4 Black Kids to encourage and educate Black children about AI and machine learning at a young age, to hopefully foster a stronger presence in the field. He noted that AI lacks historical perspectives, and his goal was to include more Blacks in the process.
Lawrence's organization offers after-school support for children aged 5-19, scholarships, and academic guidance. Andres Lombana-Bermudez, a Harvard University faculty member, advocated for addressing stereotypes in AI by increasing diversity, and also diversity in thought. He suggested bringing sociologists, lawyers, political scientists, and other social science professionals into AI and ethical discussions.
He also hoped that future generations, having grown up with technology, would be better equipped to address the biases and accessibility issues that have plagued the field.
Lawrence regretfully noted that his AI projects' teams in the past 25 years lacked more than two white colleagues while he himself is black.
According to the Taulbee Surveys conducted by the Computing Research Association, more than two-thirds of the U.S.'s doctoral degrees in informatics, computer science, or information were awarded to non-U.S. citizens, and racial data was not provided for these recipients.
Sources:
In incorporating enrichment data, note that this report emphasizes fostering a diverse and inclusive AI environment across all aspects of the industry. To combat systemic racism, some essential factors include implementing diverse training data, creating inclusive development teams, promoting public awareness and education, conducting regular bias audits, establishing ethical governance, and monitoring and improving continuously. All these efforts can lead to the development of fairer and more equitable AI technology.
