Howard and Google Aim to Improve AI Tech for Black Users
Howard University and Google have launched Project Elevate Black Voices to enhance the experiences of Black users with AI. They released over 600 hours of vocal data documenting various African American English dialects to assist in correcting biases in speech recognition technology. Ownership of the data remains with Howard, but it will be accessible to other HBCUs and Google’s AI teams. Gloria Washington emphasizes the project’s goal to provide a better user experience for all dialects.
Researchers from Howard University are teaming up with Google to address a glaring issue in artificial intelligence: improving the experiences of Black users with technology like Siri and Alexa. Under their collaborative effort, known as Project Elevate Black Voices, the pair released an enormous dataset on Tuesday — over 600 hours of spoken data capturing a diverse range of African American English dialects and accents.
The goal? To assist AI developers in correcting biases that often cause automatic speech-recognition (ASR) systems to misunderstand or fail to recognize commands from Black users. The partners noted that many users have had to tweak their natural voice patterns to be understood properly, which is an unfortunate necessity in today’s tech landscape.
Howard University gets to keep ownership of this valuable data, which was gathered from community events across 32 states. However, the win doesn’t stop there — historically Black colleges and universities and Google’s AI teams can also access the findings. Howard researchers are adamant about ensuring the data is used to reflect the needs and interests of marginalized communities.
Gloria Washington, an associate professor of computer science at Howard and a co-principal investigator on the project, remarked that voice assistant technology should genuinely understand the variety of dialects. “It’s about time that we provide the best experience for all users of these technologies,” she asserted.
The researchers are considering future broader access to this extensive database but want to prioritize organizations whose work aligns with values like inclusivity and community-driven research, according to the recent news release. This is a significant step forward, not just for Black users, but for the technology as a whole, paving the way for advancements that embrace diversity in communication.
It’s evident that this project marks a crucial initiative to address the ongoing bias in AI. With hundreds of hours of data now set to inform future developments, there’s a hopeful sentiment that such strides will create a more equitable tech landscape for everyone.
Howard University and Google’s partnership aims to rectify biases in AI technology that affect Black users, as highlighted by the release of over 600 hours of vocal data. The project hopes to enable better voice recognition for diverse dialects, allowing technology to authentically serve all its users. The focus on inclusivity and community needs is paving a path towards a more equitable tech culture. Gloria Washington’s statement underscores the urgency of this initiative, emphasizing the importance of understanding various dialects and accents within voice technology.
Original Source: www.insidehighered.com
