Siri’s Second Act

[00:00:00] Isn't home where we all want to be? Reba here for Realtor.com, the Pro's number one most trusted app. Finding a home is like dating. You're searching for the one. With over 500,000 new listings every month, you can find the one today. Download the Realtor.com app because you're nearly home. Make it real with Realtor.com. Pro's number one most trusted app based on August 2025 proprietary survey. Over 500,000 new listings every month based on average new for sale and rental listings, July 2024 to June 2025.

[00:00:33] Welcome to Tech News Briefing. It's Friday, June 5th. I'm Imani Moise for The Wall Street Journal. Apple is preparing to unveil a new Siri, but will it be enough to catch up in the race for AI dominance? We're breaking down the company's latest strategy ahead of its Worldwide Developers Conference next week and whether investors will buy it. Then, an AI model has solved a math problem that's puzzled some of the world's brightest minds for 80 years.

[00:01:01] We'll dive into why the breakthrough has some experts rethinking what AI is capable of. But first, for decades, Apple has been synonymous with innovation in Silicon Valley. But when it comes to AI, the company is starting to look old school. While rivals like Google and challengers like OpenAI have spent the past few years racing to build smarter chatbots and AI assistants, Siri has become something of a punchline. Uh-huh. Nobody was talking to you, Siri.

[00:01:32] But that may be about to change. Apple is making the case that their chatbot can change the way consumers interact with AI. Kind of like how the iPhone changed how we interact with the internet. We'll get a clearer picture of that strategy on Monday during the WWDC keynote, where Apple is expected to introduce a new version of Siri built on top of Google's Gemini AI model. WSJ Apple reporter Rolf Winkler joins me now to explain.

[00:01:58] Apple is behind in this AI arms race, but the company says it still has a path forward to winning in the AI space. How is that possible? That's what's fascinating about this. As far behind as Apple is in AI, there's an argument that is made by Wall Street analysts, by some people who've left Apple. Why? Because we all still use iPhones. If OpenAI wants to reach people with chat GPT, it has to go through Apple.

[00:02:24] All of these companies long-term still today have to go through Apple. You know, as dumb as Siri is, has been the last couple years. Has anyone traded in their iPhone for an Android because it's got some better AI features that Google has come out with? There's certainly no mass defections. So, if Apple can finally create a Siri that is your guide to this AI world, this could flip.

[00:02:54] If you have an iPhone, Apple probably knows a lot about you. And having an AI agent that can access all of the personal information in your phone probably sounds exciting to some people and horrifying to others. How does that feature mesh with Apple's stance on data privacy? Well, this is where it may be an advantage and a handicap. The advantage is Apple is known for its stance on privacy. It's not selling your information to advertisers, for instance, the way Google and Facebook do. That's not Apple's business model.

[00:03:23] In fact, privacy is something they take so seriously internally, it can hamstring some of their engineers who need data to train AI models. The information you have, you're sharing with your iPhone and only with your iPhone, and even Apple doesn't see it. So, it's an advantage for Apple because they have this trust with their consumers already. We already give all our information to the iPhone.

[00:03:48] What if it could use all these things on the device without sending it to the cloud in a way that brings AI to us? A smartphone becomes your smart assistant in a way that it isn't yet. But, as I mentioned, it cuts both ways because to train models to make good AI, you need data. And Apple doesn't have as much as you think it might because it doesn't see a lot of the data that it has. So, this is a perpetual struggle.

[00:04:17] If AI agents become the primary way people interact with apps, how does that change the economics of smartphones? Apple has become a toll booth for the online world, right? When you want to download an app and sign up for a subscription, Apple takes a cut. When you do a search in the Safari browser, Apple gets a ton of money from Google. The point is, Apple has the consumers. They use its device, which means if you want to reach them, you've got to go through Apple. And they take their cut.

[00:04:47] But now, in an AI-powered world, if we abstract the apps away, if what the smartphone is is just the chatbot or if it's somebody we're just talking to and we say, Hey, Siri, book me a car going home. Maybe you're not opening the Uber app, but you're certainly using the Uber service. Uber is going to have to pay to reach those customers in a different way, potentially.

[00:05:09] So you can see how, if they re-engineer Siri, if it is the smart assistant that basically is the on-ramp to this AI-enabled world, Apple will be the toll collector at the front of that on-ramp and could make quite a bit of money. And what about Google? How does their partnership with Apple work? Google's an interesting case. Google is going to, just like they work with Apple in search, they're going to work with Apple on AI. So this new Siri that's powered by Google.

[00:05:39] Google is the back end. As much as Apple desperately needs to modernize Siri, it can't on its own. It just doesn't have the capability. So they need Google and Google technology, Gemini, to really pull Siri into the future. Maybe over time they'll have their own solution. But in the meantime, they just got to have something. And so Google's going to work with them. The economics of that are yet to be determined. But you could see them working together maybe for a long time, maybe the way they do in search.

[00:06:08] That was WSJ reporter Rolf Winkler. Will you use the new Siri? If you're a listener on Spotify, leave us a comment with what you'll outsource to an AI assistant and what you won't. Coming up, a problem that has stumped the greatest mathematicians in the world for nearly a century has a solution at last. How AI finally cracked it. That's after the break.

[00:06:29] All passengers, the Uber ride for Mark and Jamal's romantic weekend will depart in four minutes from Platform 6. Your ride comes with a rolling countryside sunset view and a table seat ideal for playing footsie beneath. Thank you for booking your tickets on Uber. Trains on Uber.

[00:06:58] I may have finally went over one of academia's most proof-hungry groups, mathematicians. Just a few weeks ago, an open AI model solved a problem that has confounded researchers for more than 80 years. It's known as the unit distance problem. And don't worry, we're not about to turn this into a math podcast. What's important to know is that some experts are calling this the first genuine AI breakthrough in advanced mathematics.

[00:07:26] And evidence that AI may be capable of making discoveries on its own. Wall Street Journal columnist Ben Cohen joins me now to explain how we got here and what it tells us about where AI might go next. I like numbers as much as the next finance nerd. But the equations at the top of your story did make my eyes glaze over a little bit. For those of us who haven't thought about algebra since college, how would you explain the unit distance problem? Like how complex is it?

[00:07:54] The actual problem is not really all that difficult. In fact, it is known as like the best known and the simplest to explain problem in high-level math. At its core, it's simply like if you imagine that you are staring on a sheet of paper and you put N as in any number of dots, N number of dots on a sheet of paper, how many pairs of those dots can be exactly one unit apart?

[00:08:21] And OpenAI's model found an arrangement that actually does much better. And it's more complicated to explain than just like a sheet of paper and a grid. In fact, it's so complicated that it can't really be visualized on a sheet of paper or on any of our screens. And it is also so complicated that mathematicians who spend their time thinking about this stuff couldn't come up with it for 80 years. But the important thing to know is that an AI model did come up with it and it was a little bit more complex than what the question sounds like.

[00:08:49] So what did humans miss when trying to solve this problem? It's a really good question in part because the answer applies far beyond math to fields that AI is coming for next. This is like a really highly counterintuitive solution to this problem. It required the AI to synthesize instead of specialize.

[00:09:11] And so AI can use its access to all the mathematical information and knowledge in the history of humanity to spot connections that humans with our own eyes cannot see. And so this particular problem required both combinatorial geometry and algebraic number theory, neither of which I know anything about. But the AI knows everything about it.

[00:09:38] So by synthesizing those two fields, it was able to come up with this solution that mere humans were not. And how did the model come up with this answer? Was there an engineer prompting it to think of it from different angles? Or did it do it more independently? It did it completely autonomously. No engineers or human intervention involved, OpenAI says. So you can actually read, if you're curious, the entire chain of thought of this model.

[00:10:05] I should warn you that it's about 75,000 words, which is about the length of the first Harry Potter book. So it's not exactly like easy reading, but you can see exactly how the model thought about this problem and every false start and path that it went down before it finally came up with this solution. So finally, it comes up with this solution and it flags it to the humans. And the humans at OpenAI who took a look, many of whom are like trained mathematicians themselves who work for schools and are currently on leave working for OpenAI.

[00:10:35] They thought it was like so impossible to believe that they kind of didn't believe it. They went looking for errors and they tried to hunt down the mistakes in this proof. And only when they couldn't find one did they show it to external mathematicians for proper verification. So it really says something that it was so surprising and kind of so inconceivable that even the people who had sort of set these models loose on this exact problem couldn't quite believe it themselves. So what has the response or reaction been like within the mathematics community?

[00:11:05] Everyone is super impressed. And these are people who are not easily impressed. I mean, I should say that like I've written a bunch of columns about mathematicians in the past few years. And these are people who are like severely allergic to hype. They require proof for basically everything in life, much less like claims about novel breakthroughs. And they would kind of rather like eat their calculators than vouch for shoddy work. So I was really surprised when this paper and proof came out.

[00:11:31] OpenAI included remarks from nine prominent mathematicians, like really big names in the field. And when they endorsed it, that's when it got my attention. Last question. What does this say about where AI is headed? It is getting much smarter in very specific fields. Math has lent itself to being AI-ified, I think, almost like coding has. It is logical and it's discreet and it's exactly the sort of thing that AI is really good at.

[00:11:58] But this is also like a solution that required creativity and intuition and the things that we don't normally associate with AI. So I think it shows that AI is obviously getting smarter. We all know that. It's also getting smarter in ways that maybe we wouldn't have expected or couldn't have imagined. And I think that is also terrifying and thrilling and all of the things all at once. That was WSJ columnist Ben Cohen. And that's it for Tech News Briefing.

[00:12:26] If you're a listener on Spotify, be sure to leave us a comment. Today's show was produced by Julie Chang. I'm your host, Imani Moise. Jessica Fenton and Mike LaValle wrote our theme music. Our supervising producer is Katie Ferguson. Our development producer is Aisha Al-Muslim. And Chris Zinsley is deputy editor of audio for The Wall Street Journal. We'll be back later this morning with TNB Tech Minute. Thanks for listening.

[00:12:48] Your team just added its 67th AI tool and also your 67th security blind spot. The good news, the Vanta agent works like a GRC engineer in the background, finding every app your team uses, scoring the risk and drafting fixes for you. Vanta is the platform used by over 16,000 fast-moving companies like Synthesia, Nandos and Granola

[00:13:16] who are shaping the future with AI and staying ahead of AI risk. Get started at Vanta.com.