How AI Will Change Personal Tech in 2025

[00:00:00] 3, 2, 1

[00:00:03] What will the world look like 10 or 20 years from now?

[00:00:07] The Wall Street Journal's Future of Everything podcast is here to give you a peek.

[00:00:10] And we can't wait to show you what's coming.

[00:00:13] Subscribe now.

[00:00:19] Welcome to Tech News Briefing.

[00:00:21] It's Monday, December 30th.

[00:00:23] I'm Danny Lewis for The Wall Street Journal.

[00:00:25] We're hearing from our reporters and columnists about some of the biggest companies, trends,

[00:00:30] and people in tech, and what could be in store for 2025.

[00:00:35] 2024 was the year that artificial intelligence became entwined with all sorts of personal tech,

[00:00:41] beyond chatbots.

[00:00:42] From smartphones to apps, tech companies are integrating AI into everyday tools in new ways,

[00:00:48] and the trend is going to continue into 2025.

[00:00:51] How will AI change everything from weather apps to fitness tools?

[00:00:55] And what happens when this tech falls into the hands of scammers?

[00:00:59] Our personal tech columnist, Nicole Nguyen, joins us to discuss.

[00:01:04] Nicole, one thing you've been looking at lately is weather apps, and specifically why

[00:01:08] they're not always accurate.

[00:01:10] But is that something that could change next year?

[00:01:12] Yes, definitely.

[00:01:14] Google's DeepMind Artificial Intelligence Lab just announced GenCast, which is an AI weatherman,

[00:01:21] essentially.

[00:01:22] It's a machine learning model that uses similar technology to the generative AI that makes

[00:01:27] images, video, music, etc., but applies it to weather forecasting.

[00:01:33] So a typical 10-day forecast is only accurate about half the time, according to NOAA, which

[00:01:37] is the big weather governmental agency here in the US.

[00:01:42] Google's machine learning-based weather model, GenCast, produce forecasts up to 15 days in advance

[00:01:48] that were at 97% better than one of these traditionally used models.

[00:01:55] And NOAA is the National Oceanic and Atmospheric Administration.

[00:01:59] How does GenCast provide more accurate forecasts?

[00:02:02] Like, how far out can it predict the weather, and how does it compare to current forecasting

[00:02:06] models?

[00:02:07] So GenCast can predict up to 15 days in advance, and that's a leap beyond current predictions.

[00:02:13] There are current predictions that go out 15 days, two weeks, essentially, but they're

[00:02:19] not nearly as accurate as GenCast.

[00:02:22] And that's because it's been fed recent historical data from the European model.

[00:02:27] There are basically two main weather models that weather apps look at.

[00:02:31] One is a European model, one is a US model.

[00:02:34] And so GenCast was fed data from this medium-range European model, and it also takes current observations,

[00:02:42] but the key to GenCast is that it runs on a Google Cloud TPU.

[00:02:48] And for those of you who don't know what a Google Cloud TPU is, it's a special AI computer

[00:02:53] chip that can crunch numbers really fast.

[00:02:56] How fast?

[00:02:57] These supercomputers, the ones that crunch the European model and the US model, take hours

[00:03:03] to complete a forecast.

[00:03:05] GenCast can make that prediction in about eight minutes.

[00:03:08] Wow, that's dramatically different.

[00:03:10] Is this something that members of the public will be able to use themselves?

[00:03:14] DeepMind has released GenCast code.

[00:03:17] It is open source, which means anyone can take a look at it and use it.

[00:03:21] Though in 2025, next year, the public will see real-time forecasts from GenCast.

[00:03:27] Google plans on publishing that in the first few weeks of 2025.

[00:03:30] All right.

[00:03:31] And aside from Google, who else is making new forecasting models?

[00:03:34] So every year, NOAA makes a prediction on how many hurricanes we will experience that season.

[00:03:41] And that allows the regions affected by hurricanes to prepare.

[00:03:44] This year, because of advancements in modeling and data capture, NOAA made its most accurate

[00:03:53] hurricane season prediction ever in 2024.

[00:03:56] And that's great stuff for 2025 because all of that data that they were able to confirm as a part of their prediction

[00:04:05] means that moving forward, we'll know even more about hurricanes than we ever did before.

[00:04:11] Coming up, AI advancements might also be making life easier for scammers.

[00:04:16] We'll discuss some ways to stay safe after the break.

[00:04:26] What then will the future reveal?

[00:04:29] There's one thing we know about the future.

[00:04:31] It's being built now.

[00:04:33] We all have a stake in the future.

[00:04:35] The future.

[00:04:35] The future.

[00:04:36] The future.

[00:04:37] And the Wall Street Journal's Future of Everything podcast is here to give you a glimpse of what's on the way.

[00:04:42] I'm Danny Lewis.

[00:04:43] Join us as we dig into how science and technology are shaping the future.

[00:04:47] For that is where you and I are going to spend the rest of our lives.

[00:04:52] Subscribe wherever you get your podcasts.

[00:05:01] Nicole, I think a lot of us are getting spam calls and texts nearly every single day.

[00:05:05] Tell me this is something that's going to get better in 2025.

[00:05:09] Things are going to get worse before they get better, probably.

[00:05:12] I'm sorry to say.

[00:05:14] Generative AI has changed a lot of things in this world.

[00:05:18] And now fraudsters are using deepfake technology to recreate your boss's voice or their likeness on video.

[00:05:27] This actually happened this past year when finance worker in Hong Kong paid out $25 million to fraudsters who use this kind of deepfake technology in what that employee thought was an internal company call.

[00:05:40] So that's a rare case.

[00:05:41] But on an everyday level, a lot of these criminals are using generative AI just to fix their language.

[00:05:48] So there are no more linguistics mistakes.

[00:05:50] And you can also ask these chatbots to spin up a legitimate looking website.

[00:05:55] What you think is your company payroll vendor is actually just sending money right to the fraudsters themselves.

[00:06:03] Oh, that's terrifying.

[00:06:04] Are there any particular AI-powered scams that people should be especially on the lookout for?

[00:06:10] Enterprise corporate victims are going to be targeted, are targeted, and will be targeted even more in 2025 because that's where the big payout is.

[00:06:20] And especially dangerous are links to these legitimate looking websites.

[00:06:25] So you'll get an email from someone posing as a colleague, or maybe you'll get a text from them, what you think is them, or a LinkedIn message.

[00:06:34] And they'll say something like, oh, we forgot to pay this vendor.

[00:06:39] Can you submit that payment via this link?

[00:06:42] And you click on that link and it looks exactly like it should.

[00:06:46] And it's coming from someone up the chain who typically approves these kinds of things.

[00:06:51] And so next year, we'll have to be even more vigilant and more skeptical of the type of messages we receive.

[00:06:57] Are there any ways that people can protect themselves from scammers like this?

[00:07:01] Folks are going to need a safe word with their boss.

[00:07:04] Is this really you?

[00:07:06] Yes.

[00:07:06] Maybe asking identifying questions if someone's sending you a message or contacting that employee, that manager directly via phone is one way to combat this.

[00:07:17] But I always tell folks, strengthen all of your passwords, especially your most sensitive ones.

[00:07:23] So your bank account, your corporate account, your email account.

[00:07:26] If you don't have a password manager yet, it is a great time to get started with one.

[00:07:31] Gotcha.

[00:07:31] And people might be starting to think about New Year's resolutions and getting healthy.

[00:07:36] Is there new tech coming out in the next year that could help people with those goals?

[00:07:40] The latest buzzword in the health tech world is healthspan, also known as biological age.

[00:07:47] So you're probably familiar with your lifespan, which is how long you will live.

[00:07:50] Your healthspan refers to how long you will remain healthy, free of diseases.

[00:07:55] And one measure of this is your biological age.

[00:07:59] So you may be 60, but you have a healthy diet and a regular workout regimen.

[00:08:06] So actually, your biological age is closer to maybe a 45-year-old.

[00:08:11] That's a best-case scenario for anyone who's 60.

[00:08:14] And more and more technology is measuring this age.

[00:08:17] Wearables such as Aura and Fitbit have a cardiovascular age estimate that's based on heart rate metrics.

[00:08:23] But there are lots of other fun tools, too.

[00:08:27] Novo Labs has an AI face age test where you upload your selfie.

[00:08:31] This app called Death Clock next year will have this tool that will show you what you will look like at 70 with good habits and with bad habits.

[00:08:42] You get two photos, two different scenarios.

[00:08:45] So what are some trends that we might see from the longevity business in the year ahead?

[00:08:49] The longevity business is booming, and there are many facets of it.

[00:08:52] Another trend is continuous glucose monitoring for those who do not have diabetes.

[00:08:57] That's that it may look like a sticker, like stuck on your skin, and it's continuing to measure your blood sugar levels.

[00:09:04] The longevity community has adopted this med tech for those who do not have diabetes to identify elevated blood sugar.

[00:09:13] And that's because blood sugar spikes are linked to chronic issues like inflammation and vascular damage.

[00:09:19] And so there are a lot of new tools now that help you measure your blood sugar so that you can manage and regulate it.

[00:09:25] One is an app called January AI, which is going to be showcased at this year's CES and is an innovation honoree.

[00:09:33] And that lets you take photos of what you're eating and uses generative AI to predict what your blood sugar will be based on what you are eating.

[00:09:42] And there's also a needle-based sensor called Stello, which is from a company called Dexcom.

[00:09:50] And they just released a new platform powered by Google Gemini that gives you insights on how your blood sugar level is connected to your sleep, your eating habits, and other aspects of your lifestyle.

[00:10:03] That was our personal tech columnist, Nicole Nguyen.

[00:10:07] And that's it for Tech News Briefing.

[00:10:09] Today's show was produced by Julie Chang with supervising producer Catherine Millsop.

[00:10:14] We'll be back this afternoon with TNB Tech Minute.

[00:10:17] I'm Danny Lewis for The Wall Street Journal.

[00:10:19] Thanks for listening.