Why people should not dislike ads
When people usually decry the advertisement business model, it’s usually from a place of a loss of individuality. Strolling around a place of business and discovering an item on one of the shelves used to be the standard experience but now we can be told what to buy before we realize it.
Imagine you are a regular at a local cafe where the staff know you. One day you walk in and the barista behind the counter recognizes you, prepares your usual order, saves a few minutes of polite remarks, and gets you on your way faster than your average time. While you could always ask about the weather before you walk out, the morning when you slept through an alarm and are in a hurry to get to the office makes this just what you were looking for.
Or, perhaps you are running a lemonade stand in your neighborhood. As a person walks up to ask what you have in stock, you may crack a joke related to something they’re interested in. Sometimes, this element of personalization may help sell a drink but there’s never a consideration of this as “stripping the customer of their individuality” just because you said a joke that’s funny to them or any other person who likes a specific recent movie.
Marketing and advertising are fundamentally the arts of capturing and inspiring consumers. In order to inspire or inform a potential customer, one must be able to communicate to them somewhere they are already familiar with. It’s often more effective to market via product placement in someone’s favorite movie than to send an email in a language they don’t understand.
Yes, on the matter of anonymity, the notion that “you shouldn’t care about privacy if you have nothing to hide” is invalid. In a world where we already see the effects of targeted correlation (for a specific example, the majority of the billboards besides highways in silicon valley are for B2B SaaS companies), seeing pattern recognitions happen in real time should not come across as jarring.
How advertisements came about
In the old web (yes, the one before Web 3.0, the one at the start of the 21st century), people could make money digitally in one of two ways:
- From people paying for the service they’re using
- Subscriptions or paying per seat
- From allowing others to inspire users to act within their platforms
- Advertising by paying an amount per exposure, an amount per number of clicks, etc.
The former case is easier to see and understand: you have something I’m willing to give money for and I have the money you’re willing to give it in exchange for. Sure, we can base of the foundation that money comes from blood rather than barter or remark that “the price of a good is what a customer cannot get more value out of for the same amount of money”. However, this doesn’t map to products that are offered for free.
The two basic scenarios in which something gets provided by free by a company or institution are when it’s a non-profit or when there are advertisements. Most non-profits operate where and when private markets won’t provide; this can be due to lack of competency or lack of interest. The way these typically operate is by giving something for free to their target audience and showing results to donors to then yield further donations.
Advertisements solve two problems in the old web: making something accessible to people who’d otherwise not afford paying for it, making something accessible to a wide number of people. In the first case, this applies to social media and streaming services since the game is to retain attention. In the second case, this matters since the only truly large organizations in which people regularly pay for membership are governments or religions.
Something being free means its access is more democratized meaning it’s more available to bring use and value to more people. Unfortunately, there’s one discrepancy between how ads can work on the old web and how they can work in a world where AIs are prevalent - ads need to inspire an action from someone holding purchasing power.
Where advertisements work
Regardless of where the advertisement shows up:
- In the middle of a social media feed
- On a billboard along a highway
- At the start of a film in a movie theater
- Between shows on a cable channel
- Up on the ceiling of a bus or train
The action inspired needs to be either yours (the one viewing the ad) or someone you can champion into takeing an action (you’re a kid noticing a poster for a new toy so you ask your parents to buy it, you come across a venue for a date idea to suggest your significant other).
When people are the predominate user of applications and services, this makes sense. Suppose you have an online business where you sell hats, you’d like to present your store to people so they can click to your domain and buy a hat. You wouldn’t want to present your store to the penguin exhibit at the zoo.
If the world were to go in a direction where we let AI act more on our behalf, then that means AI would be doing shopping or professional services in our place. As they manuveur around, they would be given the permission and capability to go learn about a new good and attain it for us. Similar to how the rich already have personal assistants to take care of their emails or scheduling for them, ordinary people would get to enjoy some burdens lifted off their hands.
Nearly a decade ago, we’ve had a fair share of stories where AI behaves unexpectedly such as when a San Diego news anchor got everyone with an Alexa a dollhouse. So have we gotten more confident with letting “agents” do things on our behalf?
Really, I don’t think this is the right question to ask at all. Just because you train a pet to be potty trained doesn’t mean one day you hand it your credit card.
Am I saying AI is nothing more than a pet? No. What I’m trying to point out is that our relationship with AI is not going to be identical with our relationship to other humans but it may be treated that way at times (akin to when a pet owner treats their companion as being anthropomorphized).
Advertising with AI
It would be amiss to not bring up Amp in the conversation around ads supported AI. They already have companies they have partnered with to support the ads:
With a model that doesn’t require accruing new data for training as of the time of writing this.
Update (2025-10-22): Amp Free no longer requires you to opt into sharing your data for training. We changed this one week after launch because ads alone cover our costs. We have switched training mode off for any users who had previously enabled it to use Amp Free.
This, of course, being the way they went about it since there isn’t a modern equivalent to Manifold. For those who don’t know, Manifold was a company from years ago that’d allow devtools to provide their API and help it be easier for PaaS providers to integrate 3rd party offerings as part of their marketplace offering.
While Amp’s thesis around what developers would be interested in:
We’re betting that millions of devs will prefer ads to paying $20-$200+/month. We’ll see.
Seems to be playing out, it lacks the “magical” experience of someone able to browse and traverse the information highway endlessly. What made the internet great was being able to log on and see entire worlds or even connecting with the rest of the blue dot we sit on.
However, they’re taking the approach that appeals to the human via ads “displayed unobtrusively within the editor extensions and CLI” rather than ads for the agent to pick up and make decisions around. This overall makes sense, a developer tool made free by informing the user of other services they can eventually incorporate into their project[s].
But what will the consumer equivalent look like?
Where ads will go
Ultimately, ads will go where users are. If the settled UI/UX of the future are different variations of the same chat UI, then we will expecting ads on the top, left, right, and bottom of the text output like how ads surround an article on an online publication. If folks are sitting in a terminal or IDE firing off commands to coding bots, then the ads will get inserted through out the UI they happen to be sitting in front of.
Most importantly, ads have to come off as not being ads in the first place. Sponsored posts in the middle of a feed are technically sponsored but they fit in like posts from accounts you’re actually subscribed to. Product placement in movies or shows is about seamlessly integrating a sponsored product into a scene in front of the viewer. Boxes containing a “click on me” CTA on the edge of a page served by the New York Times appear like another article link.
What the question of how ads will look in the future is really a question of what products of the future will look like. Chernobyl left it impossible to widely celebrate nuclear energy and the Alexa dollhouse may prevent people from wholly letting agents purchase on their behalf with no human involvement. Instead of a cumbersome AI that feels more like an uneducated intern, the successful ads model of the future is going to be one where ads ideally feel like “hidden gems” more than a “necessary tax”.
If you’re asking “can an ad go here instead” or “will people depart if I start introducing ads”, then odds are your consumers may lead astray upon adding them. If you’re asking “my users want more than I can offer them, what else can I suggest”, then you’re more on the right track. Ideally, you want your users to be in a point where they’d be at least curious what ads you’d tell them about.