EP 458: AI News That Matters - February 10th, 2025

**SPEAKER_1** (0:01)
This is the Everyday AI Show, the Everyday Podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life.

**Jordan Wilson** (0:17)
Because your company get fined $100 million for using DeepSeek and other Chinese AI, large language models. Is OpenAI coming for not just Google, but also Salesforce? And what's new in Gemini 2.0? We went straight to the source. All right, there's obviously a lot going on in the world of AI, and you can spend hours every single day trying to keep up with it and how it's going to impact your company and your career. Or you can do the smart thing, let us do that work for you. Most Mondays, this is our show. We bring you the AI news that matters. And there's a lot of AI news that matters this week. What's going on y'all? My name is Jordan Wilson and I'm the host. This is Everyday AI. This is your daily live stream podcast and free daily newsletter helping us all, not just keep up with AI, but how we can be the smartest person in AI and use it to grow our companies and our careers. So if that sounds like what you're trying to do, you're in the right place. The other right place for you is youreverydayai.com.
That's our website. You can go there, sign up for our free daily newsletter. Each day, we recap our podcast episode of the day, bringing you exclusive insights as well as keeping you up to date with literally everything else that's going on in the AI world. Also on our website, you can find more than 400 and I don't know, 440 now back episodes of the show. You can go listen to them all for free. Watch the YouTube videos, everything on the site. Also, I got to call out one more thing on the site because it is highly relevant. Our 2025 AI Predictions and Roadmap Series. So go check those out. It's on our website, episodes 443 to 447 All right, enough chit chat. Let's get straight into it and talk about the AI news for the week of February 10th. Hey, good morning to everyone joining us on the live stream podcast audience. We do this live almost every single weekday anyways, 7:30 a.m. Central Standard Time. So if you ever want to come join us, follow us on YouTube, I guess Twitter, LinkedIn, all that, and join cool people like Michael here and Big Bogey Face, Michelle, Sandra, Marie, Christopher, everyone, Douglas, Joe, Shane. Thanks for joining us, everyone. Let's get into the AI news. First, OpenAI is reportedly developing its own AI chip to reduce its dependency on NVIDIA. So OpenAI is advancing its plan to lessen reliance on NVIDIA by developing its first in-house AI chip, according to a report by Reuters. So the chip design is nearing completion and will be sent for fabrication to the Taiwan Semiconductor Manufacturing Company, or TSMC, using advanced 3-nanometer technology. The move is significant as it could enhance OpenAI's negotiating power with other chip suppliers and also reduce costs associated with NVIDIA's chips, which currently dominate the market with, I mean, depending on what you look at, anywhere between an 80 to 90% share. So if successful, OpenAI plans to mass produce the chip by 2026, marking a swift progression in a field where other tech giants like Microsoft and Meta have faced similar challenges. So the chip aims to initially run AI models on a limited scale with potential for broader capabilities in future iterations. So this initiative is part of a broader trend where major tech companies are exploring alternatives to NVIDIA due to rising costs and dependence on just one provider. Yeah, so a lot of the big tech companies are either trying to do it themselves, right? Because they're like, all right, we're paying billions of dollars to NVIDIA. Maybe we could just do it better. Maybe we could create our own chips. So, you know, Google and Amazon slash AWS, Microsoft have gone down that route or tried to go down. And a lot of them are still using a combination of their own in-house chips versus NVIDIA. It's not just cost. It's also supply, right? So sometimes, you know, you'll have a single provider go by, I don't know, 100,000 of NVIDIA's, you know, most powerful chips. And it can be hard to kind of stockpile your supply. And you might be wondering, all right, Jordan, like, why does this matter? Well, the entire generative AI like movement so far has primarily been powered by NVIDIA, right? All of the AI that you probably use is probably powered by NVIDIA. So a lot of the big tech companies, like we talked about Meta, Microsoft, OpenAI, et cetera, are trying to at least diversify the chips they use to power their large language models so they can scale faster, so they can have more control over their kind of uptime versus their downtime. So the strategic move by OpenAI reflects that growing trend among tech companies to develop their own proprietary technology or to have kind of a secondary partner in crime. All right, next piece of AI news. Google's Gemini 2 family has fully rolled out. So Google just released its updated Gemini 2 family of models, marking a significant advancement in their AI technology. So we actually had Logan Kilpatrick, the Senior Product Manager at Google DeepMind on the show Friday, bringing you some exclusive insights into not just what's new in Gemini 2.0, but also kind of Google's future plans around the models. So Gemini 2 includes three kind of new or updated varieties. One of them is actually a new category. So the upgraded 2 Flash, the new, even more cost effective Flash Lite, and the highly capable, the big boy finally got updated, Gemini 2 Pro. So many of these updates are now available inside of Google's AI Studio, Google Vertex for developers, and many of them are actually rolled out to the front end Gemini chat. Bless up, right? Y'all have heard me belly aching, probably in 2022, 2023 about Google kept some of its more powerful and most updated model varieties, not in the front end of Gemini chat, which I think actually cost them. So it was cool to hear from Logan on Friday, kind of their plans to bolster Gemini chat and a lot of these new models and some even new features exclusive to just the front end of Gemini were rolled out on Friday. So Logan highlighted the potential for Gemini 2 to unlock new use cases and improve productivity through its enhanced reasoning and agentic capabilities, which allow AI to perform complex tasks and assist users more proactively. Our conversation also touched on the future of work, emphasizing the role of AI in automating tasks and enabling individuals to create software applications more efficiently. So yeah, if you didn't check out that conversation, we recapped it in our newsletter, but go listen to it. I highly recommend it. Episode 457, I said this on the show. I think Logan is one of the brightest minds in AI. He previously worked at OpenAI as well as a couple other AI startups. So definitely worth catching that conversation if you missed it. I mean, listen to Fred. Fred, our live stream audience just said Logan was outstanding. He was. So make sure you go check that conversation out if you didn't already. All right, could you get find millions of dollars if your business uses DeepSeek? Maybe if a new bill proposed in the US. Congress has anything to say about it, you could. That's not a misprint. Also, I don't know or actually I don't even think this bill will make it into law. But still, it is worth talking about. So a newly proposed law in the US could lead to significant fines or jail time for using the Chinese AI app DeepSeek. So the proposed legislation filed by Republican Senator Josh Hawley seeks to prevent US people from advancing AI capabilities in China. So the proposed law would ban the import of technology or intellectual property developed in China with potential penalties of up to 20 years in prison or fines of up to $1 million for individuals and $100 million for businesses. All right. That's not like when I saw this story come across, I was like, wait, is this real? And I read it and it is. Let me repeat that. So for all you people that have been like, oh, Jordan, you need to be pushing DeepSeek. And I'm like, no, absolutely not. It's a Trojan horse or it's a potential Trojan horse. This is why, right? I kid you not, y'all. I have Fortune 500 leaders reach out to me saying, hey, we made our decision based on a series of podcasts you did on certain large language models or certain demos. So I take that responsibility.