Erin Reddick, the founder and CEO of ChatBlackGPT, spoke to students and faculty at Scripps’ Balch Auditorium on Feb. 6. The platform, launched in June 2024, boasts over 10,000 users and focuses on including Black perspectives in AI responses.
“ChatBlackGPT was born from the recognition that mainstream narratives often overlook or misrepresent the experiences, histories and contributions of Black people,” the platform’s website reads. “The need for an AI tool that centers Black perspectives and provides culturally relevant insights became increasingly apparent.”
According to Reddick, ChatBlackGPT can be a valuable resource for students at the Claremont Colleges. She cited essay writing as an example, saying that an individual asking ChatBlackGPT to generate an essay on Black history would receive a more accurate response than an individual using ChatGPT.
She added that the original ChatGPT, when given this prompt, fails to mention the KKK and other significant events in Black history that still carry consequences today.
Bringing up another example, Reddick — who is currently pregnant — said that ChatBlackGPT answered her questions about giving birth in greater depth than ChatGPT did, offering important information about how Black women disproportionately have more C-sections than other women.
Reddick, a public interest technologist at the intersection of AI and culture, said that ChatBlackGPT is meant to serve as an alternative to ChatGPT. According to Reddick, ChatGPT is influenced by racial biases. She decided to launch the alternative platform after being laid off in a large downsizing by Meta.
“[ChatBlackGPT] is filling a gap and creating a generative AI experience that more people could relate to because [ChatGPT] wasn’t really in-depth for everyone,” Reddick said. “I think multicultural AI is important. I just started with the Black community because that’s the one I could validate.”
Multiple forms of AI have been exposed as vulnerable to racial biases in recent years. Rite Aid, Reddick pointed out, was forced to ban AI technology after continuous incidents of falsely flagging people of color and women as shoplifters.
Kyle Thompson, the director of learning at Harvey Mudd College, said that there is no such thing as unbiased data. As a professor of an ethics and AI course, Thompson said that working to mitigate bias in AI is important.
“I hope that [ChatBlackGPT] starts conversations about what AI use is generally, and it gets people to realize that when you’re sitting down in front of GPT, it’s not a representation of some kind of universal, neutral human perspective, or some neutral computational perspective,” Thomspon said. “It very much has to work within the data that it’s given and how it’s trained, and the parameters that function as it does.”
Thompson said that it’s troubling not to know which data sources an AI program is using. Without this information, people are typically unaware of potential biases, which he said could be dangerous when large corporations have stock in AI.
“Not watering down history is a public interest, especially right now when there’s classes and information being suppressed or banned,” Reddick said. “I think that creating a technology that can harbor that information and safeguard it is really important because once it’s in AI, you can’t take it out. It cannot be unlearned. It’s also a preservation effort.”
Reddick said that she hopes to partner with and support others in building similar AI platforms for their own cultures. She is also trying to incorporate intersectionality into her platform and in platforms to come.
She added that she recognizes the controversy of AI but that she also doesn’t see it going away anytime soon. To her, it only makes sense to work with it and to make peoples’ experiences with it as positive as possible.
“If you’re worried about AI … or you know the dangers of it, read up on it, become an advocate for it,” Reddick said. “Work on ways to organize and draw attention.”
Facebook Comments