Rebel Tech Newsletter: OpenAI x Whisper

published4 months ago
3 min read

February 14th, 2023

The Rebel Tech Newsletter is our safe place to critique data and tech algorithms, processes and systems. We highlight a recent data article in the news and share resources to help you dig deeper in understand how our digital world operates. DataedX Group helps data educators, scholars and practitioners learn how to make responsible data connections. We help you source remedies and interventions based on the needs of your team or organization.


"OpenAI recently open-sourced Whisper, a speech recognition and translation system trained on 680,000 hours of multilingual data from the web. It offers transcription and translation capabilities in multiple languages. The primary audience is AI researchers studying the robustness and limitations of the system. Whisper has limitations, such as the potential to include words in transcriptions that weren't actually spoken and unequal performance across languages, especially for languages with limited representation in the training data.”

With this quick read (it'll take you 3-4 minutes), there is a decisive conclusion and a brief commentary to leave you to ponder:

  1. A 19% error rate for Black men users over white men users solidify that Whisper is white people's tech. It'll perform well, e.g., be "robust", for simple multi-person conversations like legal proceedings or conferences. It wasn't built for the language texture of Black, Latin, Asian and Indigenous people. So AI community, go wild and do all the research on a platform that excludes 80%-85% of the world's population. Good luck with that!
  2. Is someone building AAVE-inclusive and culturally-responsive speech recognition platform? If not, should someone really be working on this? I'm truly on the fence on this one. On one side, building speech recognition systems with Black, Latin, Asian and Indigenous people as the centers would be more equitable. But on the other side, there are cultural phrases like "don't take no wooden nickels" that aren't for a database to house. Sometimes what's spoken can't be translated into text. It's meant only for humans to interpret. Share your thoughts with me on social using #WhisperInColor (@csdoctorsister on twitter and brandeis-marshall on LinkedIn).

Like what you're reading? Find it informative and insightful? You can sponsor the Rebel Tech Newsletter and follow on LinkedIn.


AI Ethics Weekly is a Newsletter managed by Lighthouse3. Lighthouse3 is a strategic advisory firm that helps global organizations innovate responsibly with AI & emerging technologies. The Newsletter spotlights the difficult AI ethics conversations everyone seems to avoid. Subscribe to receive the juicy knowledge every Friday.


May this serve as your reminder, sis. Don't act like you forgot...

Rihanna is blessed, unbothered and a boss. And so are you!

Did you get your BWD Summit tix?!? I don't want you to have FOMO in September when prices are $799 (in-person) and you see the swag -- yes, the YETI is coming back. That's all I'll share, for now!

Black Women in Data Summit

September 23-24, 2023

ATL | Online

Early Early Bird Tix Available NOW til Feb 15th at 11:59 EST

$399 In-Person | $79 Online

And sis, join as a BWD Annual member ($350 investment)! You'll get a complementary tickets to the BWD Summit 2023 and we'll add 3 virtual tickets to our sponsor-a-sis pool. BWD Annual also gives you free access the conversations, worksheets, resources and other BWD events. We're gonna thrive in this data industry together. Our next BWD Community event is happening February 26th and we're talking about "Why You Need a Virtual Assistant". I'm sharing all the tea I've got on how I say 'NO' and don't feel (as) guilty about it.

Daily-ish rest routine suggestion: Listen to 2-3 podcast episodes this week that feeds your soul. Here are my 3 fave podcasts: TriggeredAF, Blessed+Bossed Up, Unpolished MBA

Follow us on social



Data Conscience Book Tour | Samuel Dewitt Proctor Conference 2023

Since 2003, The Samuel DeWitt Proctor Conference, Inc. (SDPC) has become an esteemed global network of diverse ecumenical faith leaders, seminaries, denominations, and non-profit organizations engaged in the work of justice across a myriad of prescient issues. SDPC aims to treat the people’s needs as holy with vision, by faith, and through action. Engaging the epicenters of equity, economics, and communal experiences through education, advocacy, and activism.

I'll be speaking alongside Fallon Wilson, PhD and Rev. Saeed Richardson. Our session will equip church leaders to develop sustainable technology infrastructure plans mitigating the barriers surrounding technology, information access, and digital equity.

CLICK HERE to grab your copy of my book, Data Conscience: Algorithmic Siege on our Humanity!


Stay Rebel Techie,


Thanks for subscribing! If you like what you read or use it as a resource, please share the newsletter signup with three friends!

Brandeis Marshall - DataedX

Learn how to make more responsible data connections. I help educators, researchers and practitioners align data polices, practices and products for equity. Sign up for my Rebel Tech Newsletter!

Read more from Brandeis Marshall - DataedX