AI Builds Itself — And Might Be Coming for Your Job - DTNS 5204

0:02This is the Daily Tech News for Wednesday, February 11th, 2026. We tell you what you need to know, give you some important context, and try to help each other understand. Today, why one developer doesn't need to code anymore and thinks you need to prepare for the tools that will do your job. Exactly. I'm Tom Merritt for now. And I'm Sarah Lane. Let's start with what you need to know with that big story. So, OpenAI has started using its GPT 5.2 model instead of its O4 model to deliver results in deep research. Deep Research is that feature that pulls together a research report. It draws from multiple sources that it goes and finds on the web.

0:43And you can now specify what sources you want it to use. You can either ask it to just prioritize those along with others it finds or limit itself to just the ones you say. And then you can see the report as it's coming together. And now, while it's working, you can ask it to make changes like throwing in an additional source or making an edit midway through the report. You can also connect apps for it to look at as well. OpenAI says it's more accurate than ever before. This follows the release last week of OpenAI's GPT 5.3 Codex and Anthropics Opus 4.6, both of which have received a lot of attention because of the, in some cases, what people say are vast improvements in what they can do.

1:26So much so that some folks think that the models are getting too good. developer Matt Schumer's blog post, Something Big is Happening, was at the very top of Tech Meme this morning as it received a large number of comments on social media. Here's the key paragraph I think is getting a lot of the attention. I am no longer needed for the actual technical work of my job. I describe what I want built in plain English and it just appears. Not a rough draft I need to fix, the finished thing. I tell the AI what I want, walk away from my computer for hours and come back to find the work done. Done well, done better than I would have done it myself with no corrections needed.

2:06A couple of months ago, I was going back and forth with the AI, guiding it, making edits. Now I just described the outcome and leave it. He described Codex as exhibiting what he called judgment and taste in developing apps for him. And he noted that in the release note for Codex is the following paragraph. GPT 5.3 Codex is our first model that was instrumental in creating itself. the Codex team used early versions to debug its own training, manage its own deployment, and diagnose test results and evaluations. Schumer then goes on to basically sound the alarm that your tasks are about to be replaced.

2:44And he makes several recommendations for what he thinks you should do to prepare. He's not a doomer, but he says, you need to be ready for this. Chief among them is use the tools, adapt to the new ones as they come out. He makes a point of saying, if you're on the free version, you're about a year behind. So if you're like, my tools aren't that good, just wait. When they come to the free version, they will be. And he encourages people to just build things you didn't think were possible using this stuff. Also, will they come to the free version? Oh, yeah, I think they do. They generally have. Okay.

3:19Okay. I mean, as a free version person myself, some of the stuff I'm like, yeah, it sounds really good. But for those of us who just – we can't subscribe to yet one more thing month to month, I do wonder how much the – I pay$15 a month or I pay$0 a month actually changes our perspectives. Sure. And he acknowledges that. But I believe that they will continue to update the free versions at the same cadence they have simply because they want them to get better so that people use them more and then upgrade to the paid versions. Also, they're putting ads on them. At least OpenAI is, right? So they want people to think they're good enough to use as well.

4:11But what do you think of this idea that as they continue to improve, we are hitting that inflection point finally? And I thought what was interesting about Schumer is that he's not someone who's been sounding the alarm about this for years. It's somebody who said, yeah, I was with you. I didn't think they were, you know, they were helpful, but not all that great. And suddenly, now they're that great. And I'm a little shaken by that. Yeah, I, I, I think I'm, I'm with Schumer on this kind of, you know, because I think I'm sort of clinging on to like the old way of life a little bit, you know, where I'm like, you're better than me.

4:47No, I'm better. Nobody's better than me at doing this particular task. um but um but yeah i i have quite a few friends who these days are like i mean you got to get on board you know you have to use these tools so that you free yourself up for you know you know hours in the day that you didn't have before because you were doing this yourself and and and And there's a lot of like, I don't know, I feel weird about the whole thing because I'm like, no, I have a brain. I can do it. But yeah, this is our new normal. It is. And I think we're in the worst part of any technological change, which is before it has actually taken effect.

5:41Right? You may be skeptical and say, I don't know. I think Schumer is exaggerating. These tools, even the ones he's talking about, still aren't that good. But if you're not, they haven't had the effect yet. It still will take a while for people to adopt them, for regulatory agencies to adapt to them and allow them to be used. And that is always the scariest part because it's uncertain. You don't know how they're going to affect your job. You don't know what they're going to replace. You don't know what they're going to be good for you with because you haven't had enough chance to try them out and explore new things and discover the things you didn't think were possible with until these tools came along.

6:20And that's not me trying to say everything will be roses and sunshine, of course. That's just me saying like because we don't know, we tend to imagine the worst. Totally. Totally. um just last night a friend of mine was like i was in the sauna for an hour and i just had my ai agents do all my work for me and i was like what work yeah i'm curious what kind of work that was like what what were you doing and he was like oh you know just you know stuff um and that okay and that is i mean he's he's a you know very uh he's he's he's he's a cool dude but um that kind of stuff more and more is just going to be part of life.

7:02You can say, nope, don't want it to be part of my life. And that's, that's fine. Um, I'm kind of on that side at, you know, right now. Cause I'm like, no, no, you know, nobody's smarter than me, but, um, but yeah, that's, that's where we're going. Yeah. I'm, I'm more on the side of if it can help me do what I want to do better, then that's fine. I don't worry about it replacing me until it can become me. And I guess I shouldn't dismiss the idea that eventually it'll get good enough to actually become me. But I more see it as like, oh, can it help me achieve what I want to do faster? When it can, great.

7:46I'll use it. When it can't, I won't. And I have not seen it able to write things the way I want them to write. What Schumer is saying is like, yeah, that's because everybody focused on code at these companies, but he thinks as they start to expand what these tools focus on, that it'll get good enough. And that made me wonder like, okay, if this tool gets to where it can write the DTNS rundown and I look at it and go, yeah, that's exactly what I would have written. You made all the right choices. Do I get upset about that? Or do I say, okay, great. Now I have more time to internalize stuff, to learn about stuff, to come up with insights that it would not have.

8:26It's like, how much do you want this to be like your project versus, oh, that actually gave me two hours. Yeah, yeah. That I didn't have before. Because these things are not at the point, and I'm not saying they never will get there, but they are not at the point to provide insight and instinct and introspection. There are still things that the human brain could do that these can't. And I think that's where we're able to assert humanity's value add to things and continue to work and provide satisfaction to ourselves. Yeah, they're doing daily tech headlines, daily tech news show, all the things, cybersecurity headlines as well.

9:09I will often run things through chat GPT. It's sometimes flawed, but usually chat GPT. And, you know, I look at it and I'm like, yeah, I mean, the information is there, but it doesn't sound like me. I just would never have written that, you know? And so I use it kind of as like a barometer of like, okay, like what's more human versus less human? And I think we are still in that phase of these tools being extremely helpful to a lot of people for a lot of reasons. But do they need to replicate you? And if they don't replicate you, then you don't want them. Yeah. And I don't think they can entirely replicate it.

9:58I think that gets lost as like, oh, well, they're good at language or they're good at coding or they're good at these things or they're smarter than humans. And so they replicate us. And I'm like, I don't think they replicate every single aspect of humanity. So, you know, that is the area to focus on. I just want to reassure people that we are both humans. Yes, we are. I mean, you know, I could scratch myself right now and, you know, draw blood. I won't. I'm not going to do that, but I could. No. Because I'm a human. We don't need to do that because we're supported by humans. We are. We are. Speaking of support, DTNS is made possible by all of you listening right now.

10:44Thank you to all of you. And especially thanks to Miranda Janelle, That Charlie Dude, and Justin Zeller.

10:53There's more we need to know today. Let's get to the briefs. Let's do it. Razer has already sold out of 1 ,337 limited edition 20th anniversary Razer BoomSlang mouse models. Okay, so 20 years ago, the beloved BoomSlang was one of the first to offer a precise tracking wheel. People loved it. The new one gives you a 45 ,000 DPI optical sensor fourth gen optical mouse switches to prevent double clicking and debounce delays, 8 ,000 hertz hyperpalling for lower latency, and PU leather buttons. It also has an RGB lighting that can react to chroma integrated games. Buyers also get a dock for magnetic wireless charging and LED display frames.

11:49Again,$1 ,337, but it sold out. Yeah, I did the math. If you sell 1 ,337 of these at that price, that's$1 ,787 ,569. Oh my goodness. I got to love it. I got to love a little throwback 20th anniversary kind of thing. With the Razor crew, you know who you are out there. That's awesome. And yeah, we're lead haxers. Yeah, elite price for elite budget, right? Discord continues to deal with blowback against its policy of age verification. Thanks to RWNash on our subreddit for helping us cut through some of the clutter and find some good information about it. Discord clarified on X that it is not requiring everyone to complete a face scan or upload an ID to use Discord.

12:52And they are also not going to require everyone to do age verification to use Discord at all. So this is, it originally sounded like everyone would be restricted from adult content without doing a face scan or ID upload, but Discord says that is not the case. They wrote, for the majority of adult users, we will be able to confirm your age group using information we already have. We use age prediction to determine with high confidence when a user is an adult. This allows many adults to access age-appropriate features without completing an explicit age check. So they use some machine learning to say, nah, we're pretty sure that's not a child.

13:32We will not require you to do age confirmation. So they are saying, most of you will still be able to do whatever you do on Discord because most people don't even come across age-restricted content. But when you do, if you're an adult, the machine learning will be able to tell. And in the few cases where it can't, that's when you do the face ID. The face scan never leaves your device. IDs are deleted after age confirmation. And Discord says it will only receive your age, not your identity. Yeah. So I know there were some folks in upheaval over the idea of Discord making you prove who you are and how old you are.

14:15And as you said, that is not the case. I do find it odd that Discord's like, we already kind of know how old you are. Kind of going back to the AI sniffing out stuff, it's like, well, you probably do know how old I am for the most part, but depending on how much you use Discord and what servers you're on at the time, I would like to know more about that. Yeah. And they're not saying they're using machine learning to determine your age. They're using machine learning to tell if you're close to 18 or not. So that is a pretty blunt instrument. and it's based on what servers you're on, what channels you access.

15:08I'd be curious to know if they're looking at what you write, if they're doing semantic analysis on that, that would be interesting to know. But there's some good work out there that machine learning can just kind of look at mouse movements sometimes and make a good guess at what age range you're in. I would be curious to know more about that as well, but I much prefer that to say, like the machine learning will let us know if we need to look at someone because it looks like maybe they're close to 18 and they're trying to access age-restricted stuff. And the rest of the time, we just ignore everything.

15:39Totally. Yeah. Well, Google is adding a way for you to buy stuff when in AI mode on search. The feature is called direct offers. Stores can offer discounts and let you buy items from Etsy and Wayfair from within AI mode. Google's also testing sponsored listings in AI mode as well. Yeah. So these are ads, right? Is that what you'd call this? Yeah. Okay. Direct offers, aka ads. I know that this deep mind said we're not going to put ads in Gemini and they have not actually put ads in the Gemini product, but AI mode is Gemini. So you'll see a lot of people saying these are going into Gemini because technically Gemini is all these things.

16:27And I don't think anyone really believe that Google would just not try to make money off of ads when it's the largest ad company on the planet. So I don't know if I mind this or not, but this is how Google continues to grow, I suppose. Yeah, this doesn't really move the needle for me. Knowing that Google does Google does. And that's how it makes money. This doesn't upset me in any way. I don't really know. I don't know. The whole kind of like, you know, shopping within an experience type thing. Google is not alone. You know, as the company who's doing this. I don't know. The jury's out, I guess I would say at least for my household, like how much does this actually make us buy things?

17:26I'm not sure yet. Yeah. It'll take a while to get trust that this will work properly when I'm in AI mode. Like, would you like to buy it from Shopify and me say yes. And it just makes it happen. Right. Like, which would, which would be very seamless, right? Like that's the idea, but how much will you say? Yeah, I want that. And I guess I'll try it a couple of times. And if it works, then I'll get used to it and it'll be kind of great. There you go. University of California, Berkeley's Annalie Saxenian and entrepreneur Vivek Wadhwa studied 596 Indian high-tech startups established between 2016 and 23 and found that overseas experience plays a diminished role in the success of Indian startups.

18:13AI, fintech, mobility, and enterprise software companies did better financially when they were led by domestic founders. Now, this is counter to the narrative. The narrative has been you send somebody overseas, they get experience, probably in the U.S., maybe in Europe, maybe even in China. They come back to India with all of these expertise, and they help their startup do better. Turns out professionals with overseas experience only did better in specialized roles. The authors believe that knowledge of the local market is proving more valuable. So somebody who's stuck around in India and knows what the people in their region need and want and desire do better with their startups.

18:52This comes as Rest of World is reporting on the first AI Impact Summit happening in India, which sees companies from the global south positioning themselves as great places for investment for AI companies. India has a wealth of data and the most AI use per capita in the world. Irwanda, or maybe not per capita, probably the most AI use in the world because they just have a huge population. Irwanda and Nigeria are attractive as sites for scaling hubs. And the United Arab Emirates pitching itself as attractive place to raise money to do your infrastructure build out. Shocker. Yeah. All of this just really interesting as looking at the global south and India in particular as sort of a rising hub in a way that I don't think people have thought about India.

19:38I think people think about India as a place for outsource stuff or an alternative to China to locate your factory, maybe. But it may be becoming a center of industry. In fact, it is becoming a center of industry all itself. Well, I think, you know, India is very interesting specifically because, yeah, like you said, it's a huge market. You know, a lot of people live in India. And Indian culture is, you know, it depends on where you live and who you are. But, you know, it's very nationalistic, I guess, would be a way to describe that. And, yeah, the idea that like, oh, we outsource things to India, but that doesn't come back to us, you know, as our product is very short-sighted.

20:29Um, you know, the Indian market is, you know, more and more, you know, growing as, uh, you know, uh, it, it, it's not only, uh, you know, people having more money, but, um, just, you know, that, that huge swath of people, humans on the planet who are in the industry, you know, interested in tech, but like tech in certain ways. um that is you can't ignore that yeah that that's that's kind of that that's where i see things i think it's fascinating to to see the shift of of a company and and you may be smart to go hire someone from india if you're operating outside of india especially if you want to break into the indian market but even if you don't uh you you may find that you you're not sending people from India elsewhere to learn.

21:26You are sending people to India to learn. I think that's an interesting way to watch it. Well, and so much of, I don't know, you know, social media that we, you know, we all consume these days, you know, it changes by the day. But there are certain things where, you know, and we've all seen this, like where you kind of go like, I don't get this joke, or I don't understand what's going on here. Why are all these people talking about this specific thing? That's because you're not familiar with the market where a lot of people actually live and work and think. And so that's something I think we should all think about more often.

22:13Yeah, I agree. And share our experiences with other people as well, our direct experiences, in fact, with the gadgets we use. Oh, indeed, Tom. If you want some honest reviews from people like myself, who actually buy and live with tech products, services, what have you, then you need to join Live With It. This is a weekly look, which I host, but I'm with Tom and Roger and all of the DTNS crew each week to talk about stuff that we have spent months living with. And we have a really good idea of whether it's good for us, whether it might be good for you, you know, and, you know, and, and either or, because that is, that's the idea.

23:05It is, it is, it's a, it's a very, a show where we, we just, you know, we leave it all on the table and, and we talk about the good and the bad. Listen to Live With It wherever fine podcasts are traded or watch at youtube.com slash daily tech news show.

23:29All right, let's get to some quick headlines, kind of things that are good to know, make you feel smarter, make you look smarter. Samsung has sent out its invites to its next Galaxy Unpacked event to take place at 1 p.m. Eastern in San Francisco on February 26th. Expect to hear about new Galaxy S26 phones. We also got a minor update, Android 16 QPR3 Beta 2.1, rolling out two beta testers on the Pixel phones. Uh, that's not that big of an update, but it gets us that much closer to the official release of Android 16. TikTok expanded the launch of its local feed. So that's content related to your current location to the U S on Wednesday for people who turn on precise location sharing.

24:15It is, it is often it's been available in Europe since December. T-Bobile announced a beta coming this spring for live translation on its 5g network, meaning you don't need an app. You just tap eight, seven during a call to translate between 50 languages. You do have to register for the beta to sign up for it. TikTok, Snap, and Meta will be the first companies to be graded for their teen safety efforts by the Mental Health Coalition. Discord, YouTube, Pinterest, Roblox, and Twitch have also agreed to participate. Nintendo's Pokemon Pocopia comes out on March 5th. It's a very buzzy title. If you want to get your hands on it, they're doing demos this weekend at the Pokemon European International Championships taking place in London's Excel Exhibition Center.

25:08Google is bringing its Fitbit personal health coach preview to international premium users, as well as expanding from Android to iOS. Another buzzy game, Mugenics, spelled M-E-W. you. It's a cartoon style game about genetically manipulating a team of mutant cats to fight adventures. I love it already. Sold more than 150 ,000 copies on Steam in its first six hours. Amazon received clearance from the US FCC to launch 4 ,504 of its second gen LEO satellites to deliver internet from space. Amazon, I would say Amazon catching up with Starlink. but I was about to make that joke too. Reuters sources say bike dance is talking with Samsung about manufacturing bike dance design chips for its generative models.

26:02They aim to produce a hundred thousand units this year. And we like to end every episode of DTNS with some shared perspectives. Today, Ari has a question about the model that's helping doctors diagnose brain conditions. Yeah. Ari wrote, hi Tom and friends, regarding the February 9th discussion about brainiac, brain IAC, assisting doctors to make diagnosis. It got me thinking, how will insurance companies or Medicare respond? Should an LLM recognize a pattern that a human doctor does not? And that doctor subsequently orders tests they might not have otherwise ordered had they been unassisted by that LLM.

26:42I'm thinking of my annual mammograms and well woman screenings. They are of no cost to me because the cost of early detection and treatment outweigh the cost to my insurer of paying for potential cancer treatments? Shall extra tests influenced by LLM recommendations be free or cheap to patients a la screenings? Will we patients have to pay for them? Will doctors be penalized if they erroneously recommend unnecessary tests? Naturally, I don't expect you to know, so I'll just take this to my friend who works for a major insurer. But thanks for everything you do. I never write, but I've been listening since your buzz out loud days.

27:18Oh, wow. You know what, Ari, very, very good points that you bring up. Certainly, I think I speak for all of us when I say, you know, I'm definitely in, you know, that healthcare, you know, insurance, scratching my head at time system. And yeah, the more you know, the better, right? But how does that change what we pay for, who we listen to, and how we are healthier, hopefully, as a result? Yeah. The principle has usually been that the doctor is responsible for everything. These these diagnoses, uh, like, like the brainiac assisted, uh, are not replacing it. But I don't think Ari's wrong to say, I wonder if insurance companies will start to say, if an LLM was used, we'll have a different coverage method for it.

28:20I don't, I don't think it'll be as bad as like, if it's used, then you have to pay. Uh, but there'll be some effects. They'll, they'll account for that somehow. Well, and, and also, you know, if an LLM is used, for some sort of radiology exam, say. Is that going to be, how does that differ from what the radiologist would have done previously? Yeah. I mean, theoretically, it could actually reduce costs by catching things earlier or being more accurate, that sort of thing. But Ari, if your friend who works for a major insurer gets back to you, let us know what they say. I'd be curious too. Indeed.

29:01Thanks, Ari, for writing in. If you have thoughts on anything that we talk about on the show or anything you want us to talk about on a future show, do let us know. Share it with us at feedback at dailytechnewsshow.com. Thanks to Ari for contributing to today's show. Thank you for being along for Daily Tech News Show. You keep us in business, folks. Become a patron right now. Do it. Patreon.com slash DTNS.

29:33The DTNS family of podcasts, helping each other understand. Diamond Club hopes you have enjoyed this program.