Morality of Advertising
Ask someone if they like ads, you'll almost certainly get a "no." Nobody wants their YouTube videos interrupted by a gecko.
But directing people's attention to businesses creates real economic value. As AI assistants rapidly become our main gateway to information and the world, the question isn't about eliminating ads, it's about designing attention markets that are maximally efficient.
- Why Markets Exist
- Different Types of Markets
- Different Types of Ads
- When Markets Work
- When Markets Fail
- Ads in AI Assistants
- Towards The Efficient Attention Hypothesis
Why Markets Exist
Before talking about ads, let's think about markets. Most of economics boils down to exchanges: Alice cuts hair, and Bob needs a haircut. If they can find each other and agree on a price, everybody wins.
But those are two big "ifs":
- Liquidity: How do Alice and Bob find each other? What if Alice is off on the day Bob is free?
- Price Discovery: How do they agree on a reasonable price?
Markets exist to solve exactly these two problems: NASDAQ helps buyers find sellers, Uber helps riders find drivers, and Google Ads helps businesses find customers. They are all solving the same fundamental problem: matching supply with demand at the right price. If AI were to fulfill its utopian promise, it would have to somehow improve how markets work.
Different Types of Markets
Dealer Markets
Dealer markets are like grocery stores. They buy apples from farmers without knowing if you'll want apples this week. They hope to sell them at a profit. If no one buys their apples, they take the loss.
The dealer's job is to manage inventory risk. Dealers make it convenient for you to buy what you want, when you want it. Car dealerships, bookstores, pawn shops all work this way.
In some cases, dealers go further and become market makers, by offering to not just sell, but also buy things from anyone at any time.
Broker Markets
Brokers never touch the goods. They introduce buyer to seller and charge a fee for making this connection. No inventory risk. Real estate agents are the classic example. But also investment bankers on M&A deals, headhunters, and even matchmakers.
Brokers markets are typically used when the goods in question are non-standard, illiquid, complex, or when parties involved need confidentiality.
Auction Markets
Auction markets match buyers and sellers directly through bids and asks. The stock market, technically a "Continuous Limit Order Book," is an auction. These markets best with standardized goods and lots of buyers and sellers.
Auction markets often need a third party in the market: market makers. If you want to sell 100 Apple shares immediately, there is no guarantee that a buyer is waiting. Market makers bridge this gap by buying your shares, and holding them briefly until the buyers appear. They make money from the bid-ask spread in exchange for providing the service of bridging the temporal gap between supply and demand.
Different Types of Ads
Traditional markets have three parties involved: buyer, seller, intermediary (broker, dealer, market maker). The interaction between these parties are intermediated by money.
In advertising, there are four parties:
- Users trade attention for content.
- Creators trade content for money.
- Advertisers trade money for attention.
- Platforms sit in the middle, facilitating all three exchanges and charging a fee.
This unique structure is why we don't typically think of ads as markets, But the parallels are in fact quite obvious.
Display Ads (The Dealer Model)
These are things like billboards, Instagram story ads, or banner ads on news sites.
They work like dealer markets: Instagram doesn’t broker deals between you and Coca-Cola directly, it harvests your attention and sells it to advertisers. The platform in this case is the dealer of people’s attention. They buy your attention using the free service that they provide, and sell your attention to advertisers for money.
Like any dealer, their priority is to manage inventory (your attention) by making sure that there is always content that you would want to see on their platform. They do so by building network effects (if all your friends are posting their updates on a platform you’d want to keep checking it) and paying content creators for content.
This model is good for creating demand for new products. Nobody searched for “sustainable wool sneakers”, but AllBirds built a billion-dollar brand by buying attention through display ads.
Search Ads (The Broker Model)
Search ads are brokers. You search "plumber near me," Google instantly auctions off the top spot to local plumbers. The highest bidder wins (obviously ad systems have become very sophisticated and this is a simplification, but I don’t think it’s necessary to go into details for the purpose of this blog). Real-time auctions reveal exactly what customers are worth and the provider’s availability. An emergency plumber might bid $50 for a click at 2AM, but only $5 during the day. A plumber might bid higher prices when they are free, but stop bidding when they are booked for the next three weeks. Without ads, the search results will be mostly static, and won’t be able to match customers to the best service provider. A real time market mechanism prices supply and demand dynamically.
When Markets Work
Ad markets create value by efficiently matching supply and demand.
Display Ads Creating Value
Brand Building: Display ads enabled direct-to-consumer brands like Warby Parker and Casper to exist. Without Facebook ads, they would've struggled to reach their first customers. Once brand awareness builds, organic searches kick in, but ads provide the crucial initial lift.
Content Ecosystem: YouTubers, podcasters, bloggers exist as career options because platforms can monetize attention through ads. No ads means no free content.
Search Ads Creating Value
- Signaling Quality: Willingness to pay for clicks signals legitimacy. Scammers can't sustain ad costs when customers don't convert. That plumber paying for ads probably has happy customers.
- Real-Time Information: An overbooked plumber lowers bids. A new restaurant increases them. Search ads reflect current reality in ways organic results can't.
When Markets Fail
Ad markets can fail the same way financial markets do: monopolies, fraud, misaligned incentives. Here are some examples
- Attention monopolies: When one platform controls too much attention, they can extract maximum value from the advertisers. This creates zero-sum bidding wars between advertisers that only benefit the platform.
- Privacy Violations: There’s a fine line between personalization and creepy. Many humans can walk this line pretty well (e.g. sending your friend a personalized gift), so there is no reason to believe that AI systems cannot do this too.
- Click Fraud: Bots clicking competitor ads to waste their budget. Like spoofing in financial markets, this destroys trust and efficiency.
- Misaligned Metrics: Optimizing for clicks instead of value. A user clicks an ad, hates the product, never buys again. The platform profits but both users and advertisers lose.
Ads in AI Assistants
AI assistants (like ChatGPT) are rapidly becoming the primary gateway between users and information. What role should advertising play in this new ecosystem? Well, the same market dynamics that make ads valuable in traditional platforms still apply here:
- Funding Universal Access: Free, ad-supported models allow everyone, regardless of resources, to benefit from powerful AI tools.
- Supporting innovation: If chatGPT is how people discover new products, services, and information, an advertising system ensures new entrants can still gain visibility. Without ads, incumbents with existing brand recognition dominate search results, hurting innovation.
- Signaling Quality and Relevance: advertising on AI assistants could still serve as an effective quality signal. Businesses willing to pay for user attention likely believe in the quality and relevance of their offerings.
What might ads look like in AI personal assistants? Well, I sure hope that it won’t look like an ugly banner ad on the side of your chat. My guess is that it will resemble much more like search ads, where every time someone asks something related to some product, the AI assistant will give its usual answers plus a couple of ads based on the result of some auction mechanism. One can even imagine a system where the advertisers can use an LLM with some custom prompt to bid in these real time auctions.
There is a newer paradigm of ads, influencer advertising, where the creators sign deals with brands directly, and they incorporate their products into their content, often times explicitly in the form of a review. This is an even more natural fit for AI assistants: the ads can be incorporated into the answer naturally, with clear labels. Example scenario:
A user asks ChatGPT, "What's a good productivity app?" ChatGPT replies:
"Notion (Sponsored) is popular among productivity enthusiasts because of its flexibility. Alternatively, Todoist is a strong option if you prefer simplicity."
Personally I find this much less repulsive than other forms of ads.
Towards The Efficient Attention Hypothesis
Attention is the ultimate scarce resource we have. Platforms that capture significant human attention carry a moral responsibility to allocate it efficiently.
Advertising should not be seen as inherently good or evil. It is fundamentally a market for attention. Markets fail when incentives become misaligned, monopolies dominate, or transparency diminishes. Conversely, they succeed by effectively matching supply with demand, helping users discover valuable products, and supporting creators sustainably. Given the success that market mechanisms have had in traditional settings, one might even argue that Ads are a necessary mechanism towards a truly efficient allocation of attention.
Ads can also be viewed as an attention tax: a necessary cost that funds valuable public goods. Just as income taxes fund infrastructure, education, and healthcare, attention taxes (ads) democratize access to powerful tools like search (Google) and (maybe) AI assistants.
As AI chatbots increasingly mediate our interactions with the world, we have an opportunity to transform these assistants into skilled curators of attention. They can guide us towards genuinely relevant information while respecting our agency over attention. The critical question isn't how to eliminate ads, but how to design attention markets that genuinely serve us.