By Keith Golden Jr.
Esosa Osa was tired of seeing how the rapid growth of AI harms Black communities. She was tired of the misinformation, data centers polluting communities, facial recognition software criminalizing them and the software used to deny them opportunities. So she created Aisha AI to do better.
“The most important information tool of our generation, AI, was being handed to Black communities in a form that was actively harming them,” she said. “And so we wanted to try to create something better, that was built on Black history, Black news, Black progress that really prioritized privacy, that prioritized environmental awareness.”
Launched by the nonprofit research, education and digital innovation hub Onyx Impact as a web-based chatbot early last year, its generated responses are based on Black-led studies, history and journalism outlets. It is now being launched as a mobile app
Osa is the founder and CEO of Onyx Impact, which partnered with Chromatics AI to create Aisha.
Most AI models get their information from the internet databases, public records and historical data that reflects racial bias, studies show.
Commercial facial recognition systems misidentified and misclassified darker-skinned people 34% of the time for Black women and 0.8% for light-skinned men according to a MIT Media Lab 2018 study. A 2026 Stanford HAI study of 4 million job applications found AI discriminated against 26% of Black applicants. Roughly 40,000 more applications would have advanced to the next stage if the AI tool had recommended candidates at equal rates across racial groups.
Aisha uses sources that reflect the Black experience to inform its responses.
According to Osa, these vetted sources entail publicly available information from universities and publications like The Atlanta Voice, The Washington Informer and The Root.
“We’re not trying to choose which Black outlet is correct,” she said. “We are just showing Aisha that here’s where you should look to first, because we believe that Black journalists will have a better lens on the Black community than generic white media.”
She says that the mainstream counterparts are already determining our reality, our social media feeds, determining the news we see, who gets access to healthcare and who gets access to jobs. However, Onyx Impact is doing it in a way that is mitigating harm to marginalized communities.
They are mitigating harm in two main ways: dismantling misinformation on Black communities and having cleaner grid locations for their data centers, in Vancouver, Montreal and Toronto, Canada, which uses hydroelectricity to power its servers and returns water back to the source rather than consuming it outright. Another example she used was Aisha is designed to provide information, not keep users on the app by not being as agreeable as other AI’s.
“That’s part of the AI issue that causes AI to hallucinate more, that causes AI to lie to you right? This perpetual need to agree with the user,” she said. “And so those are the types of guardrails we put in place to help better Aisha and AI for our communities.”
Osa urges journalists using the platform to abide by the specific AI policy of their newsrooms before using the app. She recommends turning to Aisha before using ChatGPT or Gemini.
To the people who criticize her for creating this product, she reminds them that 50% of workers are already using AI and it’s going to create over $44 trillion of wealth in the next 10 years. She notes that ChatGPT pushes the death penalty if people sound Black, AI image generators default Black faces for crime and systematically reject Black job applicants. It’s here whether people want to use it or not, she said.
“If we are not creating, we are not at the table, if our engineers aren’t at the table, if we’re not decision makers in this transformative piece of technology, then people are going to use it in a way that’s actively harming us,” she said.
Be the first to comment