The Last Ad-Free Frontier

I woke up this morning, made coffee, and sat down at my balcony overlooking the tranquil Andes here in Quindio; hell, for all I know the coffee I’m drinking was grown directly across the valley in clear view every morning. You can’t blame me for smiling. Over the next few hours I researched a technical problem, drafted an email, explored some ideas for a project talking with friends, and after a particular observation, I wrote most of this blog post.

What was the observation, you ask? Not once did I see an advertisement.

No banners. No pop-ups. No “skip ad in 5 seconds.” No sponsored results masquerading as answers. No cookie consent dialogs begging me to accept tracking so advertisers can follow me around the internet like a desperate ex.

I’m paying for Claude, ChatGPT, Gemini, and a couple of other AI services. Between them, they handled most of what I needed this morning. And the experience was… clean.

Like things used to be. I’d almost forgotten what that felt like.

From Banners to Behavioral Profiles

Advertising on the internet wasn’t always this aggressive. In the beginning, it was almost quaint. A banner at the top of a webpage. A sidebar. Maybe a pop-up if the site was feeling adventurous (and you hadn’t installed a blocker yet). The deal was simple: content is free, ads pay for it. The same model that funded broadcast television for decades.

But then something shifted.

The banners multiplied. The pop-ups got sneakier. Video ads started autoplaying with sound. Tracking cookies spread across the web like a fungal network, following you from site to site, building profiles, serving you ads for things you’d already bought. The “free” internet became a surveillance economy where you weren’t the customer, you were the product.

And still, we might have tolerated it. Except the ads didn’t stay in free spaces.

Paying Customers, Still Watching Ads

The straw that broke the camel's back for most of us; we started paying but the ads didn’t stop.

Cable television was supposed to be the premium alternative to broadcast. You paid for better signal. You had the option to pay for more channels. You'd expect that money was going to those channels, so the financing paradigm changed, right? Except the ads came anyway. Now you pay $150 a month for cable and still sit through commercials like it’s 1985.

Streaming was supposed to fix this. Cut the cord, pay for Netflix, watch what you want when you want, no interruptions. That lasted maybe five years before the “ad-supported tier” appeared. Now you can pay for a streaming service and still see ads unless you pay for the premium tier. It’s a subscription to remove the ads from your subscription. Never mind how many "plates" you need for the a la carte experience.

The one that gets me... airplane seatback screens? A captive audience being served ads for credit cards and hotels while hurtling through the sky in a metal tube with nowhere to go. That's why window seats cost more... it's the only place you can look away.

And then there’s Amazon.

“By the Way…”

If you own an Alexa device, you’ve heard it. You ask for the weather, and after telling you it’s 72 and sunny, Alexa adds: “By the way, did you know you can ask me to order paper towels?”

You didn’t ask and you're not interested; that was a one-time purchase for a camping trip. Despite having been purchased for home automation and convenient timers, Amazon built a speaker that lives in your home and uses it to advertise Amazon services to you. Constantly. Relentlessly. The same company that puts ads (albeit for their own services, but still they're ads) in the AWS console, a platform businesses pay thousands of dollars a month to use, and cluttering it to the point of uselessness, figured out how to make their home assistant a persistent upsell machine.

It’s not advertising in the traditional sense, call it “suggestions” or “features", but let’s be honest about what it is: a device you paid for that is primarily asked mundane questions or simply tied to home automation is promoting the company that sold it to you, every chance it gets.

The Spy Cam Era

Way back in the early days of the internet explosion, Yahoo went through a phase. For what felt like an eternity, every other ad was for some X-something “spy cam”—a tiny camera you could hide anywhere, marketed with this unescapably creepy energy that made you wonder who exactly the target audience was.

The ads were inescapable. Pre-roll, mid-roll, sidebar, everywhere. Clearly someone was paying Yahoo a fortune to shove these things down everyone’s throats, and Yahoo was happy to take the money regardless of how distasteful the product seemed.

That’s the thing about advertising, especially at scale: the money talks, and platforms listen. It doesn’t matter if the product is creepy, or if users are annoyed, or if the experience suffers. If someone’s paying, the ads run. You’re not the customer. You’re the inventory.

The Tracking Problem

Somewhere along the way, advertising stopped being about showing you products and started being about knowing you. Every website you visit drops cookies. Every app phones home. Your browsing history, purchase history, location history, social connections—all of it feeds into profiles that follow you everywhere.

You search for running shoes once, and for the next six weeks every website you visit shows you running shoe ads. You mention a vacation destination in an email, and suddenly your social feeds are full of resort promotions. The line between coincidence and surveillance got blurry years ago.

Meta built an empire on this. Google built a bigger one. The entire economic engine of the modern internet runs on harvesting your attention and selling it to the highest bidder.

And then, quietly, something changed.

The Ad-Free Morning

Back to my morning. Research, writing, planning, problem-solving. Hours of productive work. No ads.

The AI services I’m paying for don’t need to show me ads because I’m paying for them directly. Simple enough. But there’s something deeper happening here, something that might actually threaten the advertising model that’s dominated the internet for two decades.

Think about how I used to start my day. I’d open Google to search for something or read the news, scan the results, click a link, read an article (with ads), click another link, read another article (with more ads). Every step of that journey was monetized. The search results page had ads. The articles had ads. The tracking cookies followed me to the next site, which also had ads.

Now? I ask Claude or ChatGPT a question and get an answer. Often that’s enough. I don’t click through to a website. I don’t see the ads on that website. I don’t generate the impression that funds that website (and we'll dive into that in just a minute).

Google knows this. It’s why they’ve integrated Gemini directly into search results. Half the time now, the AI summary at the top of the page is good enough that I don’t need to scroll down at all. Great for me. Terrible for every website that used to get that click.

The Revenue Problem

My son discovered the architecture of the internet last month; the servers, the stacks, the patterns emerging where scale meets data. What he didn't see was the business model underneath.

Here’s where it gets interesting, and maybe a little concerning, depending on which team you play for: the service or the product, the business or consumer.

AI models were trained on the internet. Billions of articles, blog posts, forum discussions, documentation pages; all of that content, created by people and organizations who were often funded by advertising. The AI consumed the content. It learned from the content. But it never viewed a single ad.

Now those same AI models are answering questions that used to require visiting those websites. The content creators made the content, the advertisers funded it, but increasingly the value is being captured by AI services that sat at the end of the chain and hoovered it all up.

And when users like me spend their mornings talking to AI instead of browsing the web, the ad impressions don’t happen. The clicks don’t happen. The tracking cookies don’t get set. The whole surveillance-advertising loop that funded the internet for twenty years starts to break down.

I’m not shedding tears for the advertising industry. But I do wonder what happens next.

The Bigger Disruption

The AI era will be bigger than the internet itself, or so the pitch goes from the very companies now racing to control it. If that’s even half true, then the companies building these models aren’t just creating productivity tools. They’re potentially rerouting the entire attention economy.

Did OpenAI and Anthropic see this coming? When they built products that answer questions directly, did they know they’d be disrupting the advertising model that’s funded the web for decades? Did they anticipate that every query answered in a chat window is a query that doesn’t generate ad impressions somewhere else?

Maybe. Maybe not. But you can be certain that others are paying attention now.

The signals are already there. Companies that have no business building frontier models are pouring billions into the race anyway. They’re not playing for technology. They’re playing for attention. And they see the writing on the wall.

Search-based advertising is doomed. Not tomorrow, not next year, but it's obvious where things are headed. As AI captures more of the world’s attention, advertising dollars will follow. It's easy money for recipients (but for the life of me I still don't know how advertisers can justify the mountains of money they spend).

The question isn’t whether advertising will find its way into AI. The question is how.

Will it look familiar, like the ads forced down our throats, even for paid subscribers, following the same pattern we’ve seen with cable and streaming? Or will it be something harder to see, like bias baked into the models, quietly shaping what we believe without us ever knowing?

Enshittification

There’s a term for what happens to platforms over time. Cory Doctorow calls it “enshittification”: the lifecycle where platforms start out good for users, then gradually sacrifice user experience to extract more value for business customers and shareholders, until eventually they become barely usable husks optimized purely for profit extraction.

Let’s look at the pattern.

Yahoo launched in 1994 as a directory of websites. Clean, simple, useful. Google launched in 1998 with a famously sparse homepage and organic search results. Both eventually filled their results pages with “sponsored” links at the top; paid placements that look almost identical to real results. The money was too good to resist.

Every search engine that followed took the same tried-and-proven trajectory. Start clean, gain users, inject ads.

Amazon’s search is arguably worse. I literally just searched for a Razer mouse and the second and fourth results are Logitech with a "Sponsored" disclaimer. Not because they’re more relevant, but because someone paid for that placement or because Amazon’s margins are better on those products. It’s not search. It’s storefront manipulation disguised as search.

We've seen it so many times: start with utility, gain trust, monetize attention, degrade experience. Why would AI be any different?

The Prediction

Right now, AI chatbots feel clean because they’re new and the business models are simple: you pay a subscription, you get a service. But that won’t hold forever.

The pressure is already building. These models cost enormous amounts to train and run. The companies behind them are burning cash and investors are not known for patience. And advertising is the oldest, easiest playbook in tech.

Maybe it starts small. A tiny icon next to a response indicating it’s “sponsored.” A product recommendation that just happens to favor a paying partner. A travel suggestion that prioritizes certain airlines or hotels. You probably wouldn’t even notice.

Or maybe it’s more subtle than that. Maybe it’s not an ad at all; just a slight adjustment to how the model weighs certain information. A gentle thumb on the scale. Training data curated to favor certain outcomes, certain brands, certain narratives.

And this is where it gets uncomfortable.

The Bias We Can’t See

We already have an example of what biased AI looks like: DeepSeek.

DeepSeek is genuinely impressive technology. But it’s also trained to not know certain things. Ask it about Tiananmen Square and you’ll get evasion or silence. It’s not broken; it’s designed that way. The bias is baked into the model itself, invisible unless you know to look for it.

That’s censorship driven by politics. But the mechanism is the same whether the pressure comes from a government or a corporation. If an AI can be trained to avoid discussing historical events, it can be trained to favor certain products, downplay certain competitors, or frame certain industries more favorably.

The difference between a banner ad and a biased model is that you can see the banner ad. You know it’s there. You can ignore it. But if the bias is in the training data, in the weights, in the way the model understands the world... how would you even know? It's terrifying to think about.

When a search engine shows you a sponsored result, at least there’s a label. When an AI assistant subtly steers you toward one choice over another, trained by data influenced by commercial interests, there’s no label. There’s just an answer that feels like the truth.

The Question

So here’s where we are.

I’m enjoying my ad-free mornings. The experience is genuinely better than anything the web has offered in years. For the first time in a long time, I’m using technology that feels like it’s working for me rather than trying to extract value from me.

But I’m not naive. The pattern is too consistent. The money is too big. The pressure to monetize will only grow.

The question isn’t whether AI will be influenced by commercial interests. It’s whether we’ll be able to tell when it happens.

The Answer

I wrote and edited this over the past few weeks. As I was editing the final draft to put into the queue and publish in Feb, OpenAI announced they're introducing advertising to their platform. And yesterday's Super Bowl ad from Anthropic made it very public.

AGI has arrived, but not the one we want. Instead of artificial general intelligence we're getting ad-generated income. Right there in your app, advertisements based on your queries.

The question was whether AI would be influenced by commercial interests. Turns out we didn't have to wait long for the answer.