Why Is AI Everywhere Now? (Every Tech Site Is an AI Site)

09/10/2026|,

Written by

Somewhere around a year and a half ago, late 2024 into the middle of 2025, I noticed that every technology site I read had quietly become an AI site. The phone reviews were about AI features. The security news was about AI threats. The productivity tips were AI tips. Whatever a site used to cover, it now covered AI doing that thing.

That was also when I started actually using these systems instead of reading about them. I sat down with ChatGPT to see what it could do, and I spent time understanding what Apple was trying to do with Siri. It was interesting. It was not yet alarming.

The alarming part came about nine months ago, when I realized how fast it was moving. Changes that used to take years were showing up in months. Then in weeks. I have been in IT for more than thirty years and I have never watched anything develop at that pace.

So the question in the headline is a fair one. Why is AI suddenly everywhere? I think there is one honest answer and three less flattering ones, and you need all four to make sense of what you are seeing.

First, the pace really is unusual

Artificial intelligence as a field is old. The term dates to a workshop at Dartmouth in 1956, and for most of the decades since, progress was slow enough that nobody outside the field noticed it. What is new is the kind of AI that arrived at the end of 2022, the systems you can type to in plain English.

The adoption numbers for this kind of AI are unprecedented. ChatGPT reached 100 million monthly users about two months after launch, according to a UBS analysis reported by Reuters. For comparison, TikTok took about nine months to hit that number, and Instagram took two and a half years. The UBS analysts said that in twenty years of following the internet they could not recall a faster ramp.

The web took roughly a decade to reach most American households. Smartphones took about as long. This took a winter season.

That alone would justify a lot of coverage. But it does not explain why every topic turned into an AI topic. For that, you need to understand what AI actually is, and more importantly, what it is not.

We have covered this in a few other posts, but it does deserve a review here.

AI is not a computer. It is a layer on top of computers.

This is the part I most want you to take away, because once you see it, the flood of articles makes sense.

An AI system runs on computers. Enormous numbers of them, in data centers, drawing serious amounts of power. But the AI itself is not a computer. It is software, and a strange new kind of software at that. Traditional programs are written by people, line by line, telling the machine exactly what to do. These systems are trained instead. They are shown staggering amounts of text, code, and images, and what comes out the other end is a model that can respond to things nobody explicitly programmed it to handle. Yes, they are trained, not programmed.

AI sits in a layer above the hardware and above the ordinary programs we already use. That matters because a layer behaves differently from a device. A device is a single object with a defined purpose. A layer, by contrast, can spread across many devices, services, and workflows at once. Once AI becomes a layer, it no longer belongs to one product or category. It can appear inside search, email, writing tools, operating systems, browsers, cameras, customer service, and almost every other part of digital life. That is why AI seems to be everywhere now: it is not replacing the whole stack so much as seeping through it.

Think about the World Wide Web. The web is not a new computer. It is a layer we work with on the computers we already own. Within a few years it had reached into banking, shopping, news, mail, and eventually the thermostat on the wall. Nobody bought a “web machine.” The web came to the machines they had.

AI is doing the same thing, though much faster. It reaches into anything with a software or data surface it can touch. Your phone has one. Your email has one. Your car, your camera, your spreadsheet, and the customer service line you called last week all have one. So AI shows up in all of them, and every writer who covers phones, or email, or cars, or cameras, suddenly has an AI story to write.

That is the honest reason every tech site looks the same right now. The layer landed on every beat at once. This is not a bad thing, but it is a reason.

I have watched a layer land before

At the university where I spent 23 years supporting a postgraduate school, our email ran on a system called Lotus Notes. It was a good system, genuinely. But everything about it was local. The mail lived on servers we owned, in a room we maintained, managed by people whose whole job was keeping it running.

Then smartphones arrived. Connecting Notes to a BlackBerry, and later to the early iPhones, was a nightmare. We got it working, but it fought us the entire way, because Notes was built for a world where your mail lived in a building.

Within a few months of that fight, our department moved to something new. Gmail. Cloud-hosted email that ran in a web browser. We were the first department at the university to make the move, and at the time the concept was just amazing. No server room. No mail administrators babysitting hardware. The layer was somewhere else, and it came to us through whatever device we happened to hold.

Within about a year, as I recall it, the university started moving in the same direction, because it worked. It was not perfect. But it was far better, and an entire group of people whose careers had been mail management was refocused onto other work.

That is what a layer does when it lands. It does not replace the computers. It changes what the computers are for, and it changes what the people around them do. AI is that, at a speed I would not have believed if I had not watched it.

Where the layer reached first

If you want to see the layer in action, look at the code. In my opinion, coding is the one thing AI has touched sooner and more deeply than anything else, and the clearest example is a language most people have never heard of.

COBOL is a programming language. It was designed in 1959. It still runs a startling share of the systems your money passes through. Reuters reported in 2017 that it supports about 43 percent of key U.S. banking systems and that roughly 220 billion lines of code are still in use. Nine years later it has not shrunk. A 2022 industry survey put the figure closer to 800 billion lines, roughly three times the earlier estimate, and nearly half of the people surveyed expected the amount to grow. COBOL is still out there. The people who understand it are not. They have retired, but the programs have not.

AI is now helping people manage that code. Reading it, explaining it, finding the place a change needs to go. It is not ideal. AI is not perfect, and I would not let it touch a payment system unsupervised. But when the alternative is nobody at all, it is a real answer and a good illustration of the layer reaching into something written before most of today’s programmers were born.

So “old technology” is not safe from the AI layer. The dividing line is not age. It is whether the thing has a surface AI can reach. A mainframe running 1970s code has one. A thermostat that is not connected to anything does not, and never will.

The three less flattering reasons

The layer explanation is the true one. It is not the only one, and I would be doing you a disservice if I stopped there.

Everything is being labeled AI. A feature that would have been called “smart” or “automatic” three years ago is now “AI-powered,” because that is what sells and what gets clicks. A meaningful share of the AI coverage you see is ordinary software with a new sticker on it. Marketing, plain and simple. Though this is not always good. AI is not being well received by everyone, either. In a June 2025 Pew Research survey, half of U.S. adults said the growing use of AI in daily life made them more concerned than excited, and only one in ten felt the reverse. Not everyone likes the idea of AI-powered this-and-that.

AI writes a lot of the AI articles. It is now cheap to produce a competent-looking article about anything, and the easiest thing to produce a competent-looking article about is AI. Some of the flood is the layer writing about itself. I use AI tools on this site too, and I say exactly how on the AI-Assistance Policy page, because I think you are owed that. 

I remember when using spell check was a bad thing. AI for writing can create what some call AI-Slop, but it also can help and produce good writing. AI is not evil, bad, good, or great. AI is a tool.

The money is there. Investment, advertising, and search traffic all point in the same direction right now, and publishers follow the traffic. That is not a conspiracy. It is how every technology wave has been covered, this one included.

What this means for you

You do not need to read all of it. Most of what is being published about AI this week will not matter next month. It may not even matter next week. Focus on the concepts and trends, not the details, at least for now.

What is worth understanding is the shape of the thing. AI is a layer, not a device. It is arriving on top of technology you already own, through updates you did not ask for, which is its own story. It is uneven, it is often oversold, and in a few places, like that old bank code, it is quietly doing work that nobody else was going to do or ever could do without vast amounts of money and time.

When you see the hundredth AI headline of the week, ask one question. Is this the layer reaching something new, or is this a sticker? Most of the time you will be able to tell, and once you can, the flood gets a lot easier to wade through.

If you want the plain-English version of what these systems can and cannot actually do today, start with What AI Can (and Cannot) Do in 2026.

Verified resources & documentation

Keep reading

Recent Posts

Categories