ChatGPT's explosive launch stirred up a tech revolution, fascinating and terrifying the public with the possibilities it opened up. While the company behind this groundbreaking technology, valued at up to 90 billion USD, has been highly praised for its achievements, it has also come under fire for its apparent lack of diversity within its leadership team.
After a series of turbulent events, including the dismissal and rapid reinstatement of CEO Sam Altman, OpenAI reshuffled its board. Unfortunately, this resulted in the departure of its sole female director, leaving a board composed entirely of three white, male executives. Two of these men fit the "tech-bro" mold typical of Silicon Valley, with the third being an economist known for making controversial remarks about women.
This perceived lack of diversity raises concerns as it appears at odds with OpenAI's stated mission to ensure that general artificial intelligence benefits "the entire human race." Unsurprisingly, more voices are joining the chorus questioning how OpenAI can possibly achieve this goal without including a diverse range of individuals on its board.
Even US lawmakers are beginning to raise concerns. In a recent letter addressed to Altman and the board, Representatives Emanuel Cleaver and Barbara Lee urged the company to take swift action to diversify its leadership. They argued that the lack of diversity and representation within the AI sector is closely tied to issues of bias and discrimination in AI systems.
At the heart of this debate is the concern that AI technologies developed by homogenous teams may not accurately reflect the needs and experiences of diverse communities. In this context, linguist and ethicist Margaret Mitchell, the former head of Google's Ethical AI team, has been voicing her concerns.
Mitchell, now consulting at Hugging Face, acompiler-focused AI company, expressed her concerns to CNN. She feels uncertain that OpenAI can truly develop an AI that benefits "the entire human race," as the term itself implies a narrow understanding of people's aspirations. Instead, she perceived the mission statement as reminiscent of a "white savior complex" – the belief that certain white individuals have a duty to uplift communities of color.
"If we continue to develop tech in line with the perspectives of affluent white-male Silicon Valley leaders, we're doing a good job," Mitchell said to CNN. "But I think we can do better."
Her sentiment was echoed by Dr. Joy Buolamwini, a computer scientist and activist for algorithmic justice, who criticized the reliance of AI systems as gatekeepers of opportunities. Buolamwini, the author of "Algorithms of Oppression: How Search Engines Reinforce Racism," discussed the negative impact that biased AI systems can have on underrepresented populations, including their influence on hiring decisions, insurance policy, mortgage approvals, and even medical appointments.
When AI-based tools are employed as gatekeepers, it is essential that the decision-making processes reflect the communities they affect. In light of the prevalent issue of gender and racial biases in AI, it is particularly crucial that diverse individuals are part of the development, design, and implementation of these systems.
Buolamwini's research shows that AI algorithms often incorporate and perpetuate the existing biases from their highly biased training data. Commercial language models, for example, used to generate text like ChatGPT, rely on the heavily gendered and racialized online discourse, thus amplifying and disseminating these prejudices – albeit accidentally – at an alarming scale.
OpenAI's current board members include the former Salesforce co-CEO, the ex-Twitter chairman Bret Taylor, the ex-Secretary of the Treasury Larry Summers, and the online Q&A CEO Adam D'Angelo. The Quora platform is merely "in the process of forming."
Summers, one of the current board members, previously made controversial comments implicating that inherent biological differences contribute to women's underrepresentation in STEM (science, technology, engineering, and mathematics) fields. Taylor clarified that he and the rest of the board are "unequivocally committed to inclusion and diversity." Although OpenAI has not provided a timeline for appointing new board members.
In a blog post announcing his return as CEO, Altman emphasized the importance of building a board with diverse perspectives. He also highlighted OpenAI's ongoing investments in diversity, equality, and inclusion. Despite these efforts, OpenAI's limited representation of underrepresented groups, such as women and individuals of color, persist.
Instead of limiting itself to Silicon Valley's sphere of influence, OpenAI could seek to diversify its board by casting a wider net for candidates. It could consider engaging with various communities and fostering partnerships with organizations dedicated to promoting diversity and equality, such as the Congressional Black Caucus.
Another approach is to introduce innovative recruitment strategies, including advertising board positions more broadly and actively seeking out candidates from underrepresented groups. Such measures could enrich the board's perspectives and ultimately reflect a more diverse global audience.
As Buolamwini rightly points out, diversifying the board is just one step in ensuring a fair and inclusive AI future. A holistic approach that involves partnerships with various stakeholders, robust regulatory frameworks, and diverse recruitment and representation practices is necessary to achieve a more equitable and representational AI ecosystem.
