So check it out, there’s this startup that’s trying to fix the issue of racial bias in AI. They’re called Latimer, named after this badass Black inventor Lewis Latimer from the 19th century. And get this, they’re training a new language model, right? Like a big one, to accurately represent Black history and culture. Now, the founder and CEO of Latimer, John Pasmore, got the idea when he realized that no other companies were really tackling this racial bias problem from a technical standpoint. And he’s like, “Nah, man, we gotta fix this.”
Now, here’s the deal. They’re using OpenAI’s language model as the foundation, but Latimer is training it with data from a diverse range of sources, like books and newspapers written by Black writers. They got this approach called “RAG” – retrieval-augmented generation – which basically means they can tap into the power of the language model but make it pull answers from this different set of information. It’s pretty slick.
But here’s the best part. Latimer is teaming up with Historically Black Colleges and Universities, man. They’re all about giving students access to this technology and even including them in the AI training process. That’s some next-level stuff right there. And they’re also working with this dude Molefi Kete Asante, a professor from Temple University who’s an expert on Afrocentricity. This guy has written like a hundred books, man. His knowledge is like a goldmine for Latimer.
So, let me break it down for you real quick. When somebody asks a question, like who built the pyramids, Latimer’s AI won’t give you some sketchy answer from social media or YouTube comments. Nah, man, it’s gonna pull out the facts based on its diverse training data. And they’re even looking into addressing other forms of bias, like personal bias that might have influenced an author writing a PhD dissertation.
And you know what’s crazy? This whole issue of AI bias is on the White House’s radar now. President Joe Biden signed an executive order last week, making it part of the agenda. When they asked John Pasmore about it, he’s like, “I’m more of a free market guy, man.” He’s not into the government stepping in and regulating everything. He believes market competition will naturally push companies to deliver top-notch AI products.
Now, I got some other AI news for you. Adobe, the company known for its content authenticity efforts, got caught selling fake AI stock images. Yeah, it’s nuts. These images were supposedly from the Israel-Hamas war, but turns out they were total fakes. Adobe’s gotta step up their game, man.
There’s also this Brooklyn-based AI startup called Agentio. They raised a cool $4.25 million to help YouTubers sell ads and connect them with the right advertisers. They’re using language models to analyze all this data and make the magic happen.
But listen, it’s not all smooth sailing. In California, there was a lawsuit about AI copyright, and the judge dismissed a big chunk of it. So, things can get messy sometimes.
Oh, and Microsoft just dropped their AI Copilot. It’s like an AI assistant for their software, man. You can use it with Word, Outlook, Excel, all that good stuff. It’s gonna make your life easier, trust me.
There’s a whole lot more AI news out there, man. This group of philanthropies, including Mozilla Foundation and Ford Foundation, they’re investing a whopping $200 million to mitigate the risks of AI. They wanna explore the benefits but also make sure things don’t go off the rails.
And get this, Habu, the data clean room software provider, just announced some new generative AI tools. They’ll give you insights, predict outcomes, and even generate descriptions. It’s like having your very own AI genie.
So, that’s the AI news for today, my friends. There’s always something exciting happening in this world, and I’m here to keep you in the loop. Check out the links below if you wanna dive deeper into any of these stories. Stay curious, stay hungry for knowledge, and I’ll catch you next time. Peace out.