AI Daily News March 22 2025: 💥OpenAI study finds links between ChatGPT use and loneliness 🧠Tencent unveils a Deepseek competitor 🤖Meta to let you use AI to generate Instagram comments 🔍Claude gets real-time web search 🎙️OpenAI's voice AI
14:17 Share

This is AI Unraveled, created and produced by Etienne Newman.

a software engineer, and a big soccer dad, by the way, up here in Canada. Yeah. Welcome to your personalized deep dive into all things AI, the latest happenings, all from March 22nd, 2025. You know, you send over a bunch of really fascinating updates. And what we're going to do today is, you know, really unpack what matters and connect the dots in a way that makes sense for you. Yeah. And a quick reminder, if you're digging these deep dives, you know, go ahead and hit that like and subscribe button over at Apple.

Right on. You know, one thing that jumped out at me when I was looking at this collection of news was that it's all from the same day, March 22nd, 2025. So it's like this really cool snapshot of, you know, how much is actually happening in AI, kind of like looking at a single day under a microscope.

Yeah. And as we went through it all, some pretty clear themes, you know, they started to pop out. Like we're seeing some big jumps in what these digital assistants can actually do. There's some, you know, interesting shifts in strategy from the big tech companies about how they're approaching AI. And of course, you know, we got to think about the bigger picture, right? Like

Like, how are these advancements going to change the way we live and interact with each other? Plus, you know, the competition to build the absolute best AI out there, it just seems fiercer than ever. Totally. And even just from this one day, I mean, we can pull out some massive developments. Think about Cloud AI now being able to search the web in real time or open AI, Google's

giving their voice tech like a sense of personality. And then there's the shakeup happening over at Apple with Siri and even get this, a study looking at the connection between chatbot use and loneliness. There's a ton to cover here, but every bit of it gives us a clue about where AI is headed, where you want to start. Okay, let's start by talking about how chatbots are getting a serious upgrade.

We saw some big announcements from Anthropic and OpenAI. First off, Cloud AI can now do real-time web searches. This is a big deal, right? It's a major update from Anthropic built right into their Cloud 3.7 Sonnet model. When you give a chatbot direct access to live information, it changes the whole game.

It's not just stuck with the data it was trained on anymore. It can give you answers based on what's happening right now. Exactly. So, you know, if you ask Claude about, say, the latest stock prices or some news from this morning,

In theory, it should be able to give you a totally up-to-date response by actually searching the web. And it looks like this feature is starting with paid users in the U.S., but they're planning to roll it out to free users and more countries later. But even then, you have to go into your profile settings and turn it on. What I think is really interesting here is that, you know, Anthropic is putting a lot of emphasis on making sure things are clear. The reports say that when Claude pulls info from the web, it'll give you direct citations so you know exactly where it came from.

And that's huge for building trust. It lets the user easily check if the information is legit. It's a big step towards making AI more transparent.

Right. That's a great point. So it seems like the big takeaway here is that giving Claude access to real time information, along with being transparent with the sources, makes it a real player in the AI chatbot world. It's, you know, fixing a major issue and maybe setting a new standard for how these things provide information. For sure. And along with these upgrades to conversational AI, open AI is going after something just as important.

how these AI sound and feel to us. They just launched these next gen audio models through their API. Oh, wow. This could be the start of something totally new. AI voice agents that actually have like unique personalities. I mean, imagine developers building AI that can drop the monotone and actually take on different speaking styles.

changing its tone based on what's going on in the conversation. That's pretty mind blowing. And the model they're calling GPT-4 a minute, that's the real star here. Like you can tell it to speak like a pirate and it'll change how it talks. It's crazy. Shows they're really trying to understand how we communicate. All those little details. But it's not just about sounding cool. It sounds like there are some pretty big improvements under the hood, too.

Apparently, their GPT-4O transcribed speech-to-text models are even beating their previous top-tier Whisper models in accuracy.

That's a game changer for voice interaction. Absolutely. If you want AI to work with voice, super accurate speech to text is a must. And they're smart to launch OpenAI.fm as a demo platform so everyone can try out these new voice styles. Letting developers use it through their API and the agent's SDK means they can start building this stuff into their own projects right away. Okay, so the big thing here is this potential for

way richer, way more personalized interactions with AI. Imagine AI assistants that actually have personality. But that makes you think, as AI gets better at, you know, sounding like us, acting like us, how does that affect our relationship with it?

It's kind of freaky, right? It's the million dollar question we got to figure out as a society. When AI feels more human, there are some big questions about trust, being upfront, and even how we see what a real connection even is. OK, so let's move over to some of the big moves happening with the tech giants. That news about Apple shaking things up in their AI department, especially with Siri,

That really caught my eye. Yeah, this seems like a pretty big shift in strategy. And from what I'm reading, it's because they're trying to catch up. Siri just hasn't been keeping up with the competition in the smart assistant game. Bringing in new leadership who really know AI shows they're serious about fixing things. Yeah, it seems like those big AI upgrades and personalization features they were hinting at with the iPhone 16, maybe those are hitting some snags.

And now, get this, Rockwell is reporting directly to Craig Federighi, so basically Siri isn't under John Giandrea anymore. That's a pretty big change.

And the internal reports, they're saying there were missed deadlines, problems with implementation. Seems like Apple's AI team, especially the Siri folks, they've been struggling. And the fact that AI strategy was a major topic at their big leadership summit shows just how crucial this whole thing is for them moving forward. So to me, this is Apple is putting all their chips on the table to finally make Siri a real competitor.

This shakeup could be the push they need to come up with some crazy new features. It's like they're finally admitting they need a new game plan for AI. Right. And while Apple's dealing with internal changes, Meta is making some moves of their own. Have you heard about their tests with AI-generated comments on Instagram? Whoa, that's kind of wild, right? So you can just tap a button and the AI will suggest what to say on your friend's posts.

This guy, Jonah Manzano, he saw it on X, apparently. Yeah, and it kind of makes you think about a future where AI is just woven into how we interact online, maybe making it easier for people to engage. If you're ever stuck for words, this could be a handy little feature. But it does raise some big questions about authenticity.

If all these comments are written by AI, what happens to the genuine connection on those platforms? Could make things feel really artificial. I think those are totally valid concerns. We could end up with a bunch of generic bot-like interactions.

It'll be interesting to see how people react if this goes mainstream. Can they even tell the difference between human and AI comments? And how is that going to change how we see online interaction? And then there's the news about Meta AI launching in all of Europe. That's huge. It is. Putting Meta AI into WhatsApp, Messenger and Instagram across Europe gives them access to tons of new users. It puts them right up against Google's Gemini and OpenAI's ChatGPT in one of the biggest markets out there.

And meta AI, it can do a ton, right? From answering questions to helping you write stuff, even generating images with their imagined feature. Yep. This is the AI assistant race going global. With meta jumping into the European market, I think we're going to see more features tailored to specific countries and more attention to those EU regulations. Let's talk about some of the bigger picture stuff. That study from OpenAI and MIT about chat GPT use and loneliness.

That was pretty thought provoking. Yeah, it definitely raises some red flags. Finding a link between using chat bots like ChatGPT and feeling lonely and then seeing a drop in real life social interaction. That's something we need to take seriously. Especially the part about people using ChatGPT for like personal stuff, having conversations with it. That seemed to be linked to even more loneliness, especially for those who felt like they were connecting with the AI on an emotional level.

That's a little unsettling, to be honest. And the study was done pretty rigorously to a randomized controlled trial with a good number of people. And they looked at a massive amount of data from interactions. So, you know, there's something to this.

it kind of shows that while ai can give you a way to interact it doesn't replace that deep human connection we need so the takeaway here is all about balance right ai tools are super useful but if we start relying on them too much for companionship it could mess with our well-being and how we connect with other people we can't forget how important it is to have

real human interaction in our lives. Definitely. Now switching gears to innovation and competition, Tencent announcing their new AI to take on DeepSeek. That tells you just how intense this race to build the next big AI has become. Sounds like Tencent's going all in.

Their new T1 reasoning model, they're saying it uses similar learning techniques as DeepSeq and even did better on that MMOU protest. And get this, T1 also matched DeepSeq R1 on the C Evil suite, which is all about understanding Chinese. And it even beat OpenAI's O1 model automatically.

on that test. Pretty impressive, right? Shows that AI is booming globally in different languages too. And their pricing is right there with DeepSeek R1. So I'm thinking this means we're going to see even more powerful AI applications coming out of China. Yeah. And all this competition, it'll hopefully push everyone to keep innovating.

- Exactly. And on the image generation front, those MIT researchers, they came up with a new AI tool that can make high quality images way faster than the current best methods. That's a game changer for visual content. - Wait, faster and better quality. That's a big leap. So their new take on diffusion models, that's the secret sauce for making things faster without sacrificing quality, right? - Exactly. This could be huge for creative industries. Faster and easier AI image generation means more people can use it, even on devices that aren't super powerful.

We might see creative workflows speed up and all sorts of real-time uses for this kind of AI. And just to give you an idea of how fast things are moving, we got a quick rundown of other AI news from March 21st and 22nd, 2025. Yeah, it was a whirlwind. OpenAI released their O1 Pro model, but it costs way more. Perplexity might raise a ton of money, and Google's Notebook LM got a cool new feature.

After some delays, Meta finally launched their assistant in a bunch of European countries. And Hugging Face weighed in on the White House's AI plans, saying they're all for open source AI. It shows just how much is happening in AI right now. Now let's look at how AI is being used in some more traditional areas. Governor Shapiro talking about how AI is helping Pennsylvania state workers. That was interesting.

This is a real world example of AI being used in government. They're saying it's boosting productivity, cutting down on paperwork and making services better for those state workers. It's proof that AI can really help in the public sector. And they're using it for all sorts of things, right? Like processing documents, communicating with citizens, even catching fraud.

It seems like Pennsylvania is showing other governments how to use AI to do their jobs better and serve the public. Other states and even the federal government, they might be watching what Pennsylvania is doing to see how they can modernize their services with AI. It's all about finding that sweet spot, using the power of AI, but making sure it's done ethically and responsibly. And then we saw that news about the Church of Jesus Christ of Latter-day Saints. They added an AI-powered search assistant to their big handbook.

don't see that every day it really shows you how ai is popping up everywhere even in organizations with deep traditions and established ways of doing things having an ai assistant to help people navigate a complex document like their handbook could be a huge time saver for members who need specific info or guidance so what this means is that more and more groups religious or not are realizing how helpful ai can be for accessing information and getting support it's

everywhere now. It really is. AI is becoming a part of our lives in ways we couldn't have imagined just a few years ago. This has been an awesome deep dive into just one day in the world of AI. Now, if you're getting something out of these explorations and want to help keep AI Unraveled free for everyone, please think about donating. You can find all the links in the show notes. Every little bit helps us keep digging into all this amazing AI stuff. That's right. Your

Your support means we can spend the time and energy needed to keep up with all this info and bring you the most important insights from the world of AI. Okay, so let's recap the big themes for March 22, 2025. Conversational AI is getting seriously good, with chatbots that can now search the web in real time and sound more human.

The big tech players are making big moves to up their AI game and go global. We talked about some pretty big social implications, like how using chatbots might be linked to feeling lonely and how important it is to think about ethics when we develop and use AI. And lastly, the race to create the best AI is hotter than ever, pushing innovation in tons of different areas. The landscape is changing so fast, it's important to stay informed. And on that note, here's something to think about.

We saw how AI is getting real-time web access, personalized voices, and even being used by governments and religious groups. So what do you think will be the biggest impact AI has in the next year?

And how are we as individuals and as a society going to handle this exciting but sometimes confusing world of AI that's unfolding around us? Lots to think about as we move forward. And hey, if you're trying to reach a bunch of professionals who are into technology and the future of AI, this is the place to be. Think about advertising here on AI Unraveled to connect with our awesome tech-savvy listeners.

You can find out more about advertising in the show notes. It's a great way to reach a targeted and engaged audience. And that's it for this deep dive into the world of AI. Hope you found it valuable and got you thinking. We'll be back soon to explore even more AI developments. Until then, stay curious.