Every time you say something to Alexa or Siri, or use voice to text to send a text message, you’re using artificial intelligence. While those programs can be pretty accurate, there’s plenty of times when it’s not, especially if you’re Black.
Howard University and Google are teaming up to change that through a partnership called “Project Elevate Black Voices.”
“We ultimately collected 600 hours of African American English from across the United States to improve automatic speech recognition technology,” said Dr. Lucretia Williams, a senior research scientist at Howard University within the human center AI Institute.
“So we know we use Siri, Alexa, Google Home, and sometimes, it doesn’t understand when we give it a prompt, or we tell Alexa to turn our lights on, or to turn up the music. So the goal is to make sure the data sets that these models are trained on is inclusive with African American English,” she added.
Voices were collected from African Americans around the country, from Chicago to Atlanta to New Orleans, and so many other places where dialects can be quite different.
“We spend our money to buy these technologies in order to use it. But if there are challenges that we can’t even use it with our voice … I think it’s important to know that this is a big problem,” Williams said. “There’s a large demographic that wouldn’t be able to use these technologies and are getting excluded.”
While 600 hours of voice patterns might seem like a lot of talking, in the AI world, it’s not that big at all. The data is owned by Howard University and, initially, the school will only be sharing it with other historically Black colleges or universities for their own academic work, whether it be chat bots or model evaluations or other university uses.
“We want to make sure that other HBCUs have the data set first to do research in whatever that means, to improve whatever language models they’re working on,” Williams said. “Then, in the future, the next steps is to see how we go about giving that to other companies who will want to improve their technology as well.”
It’s not clear when the data might ever get licensed to other companies to help with their speech recognition software. However, the school said once that happens, priority will go to companies whose work is aligned with Howard’s values of inclusivity, empowerment and community-driven research.
“We want to just make sure that we aren’t getting left behind in the improvement of technology,” Williams said. “We’re using it anyway. It’s here. But we just have to make sure we are thinking critically about our data, where we’re giving it to, and making sure the privacy implications are there.”
Get breaking news and daily headlines delivered to your email inbox by signing up here.
© 2025 WTOP. All Rights Reserved. This website is not intended for users located within the European Economic Area.
