ChatGPT’s $1 Billion Ad Ambition Tests the Attention Economy

ChatGPT is on track to generate $1 billion in annual ad revenue, and more than 100,000 advertisers are running campaigns on the platform, according to Reuters. Joe Marchese and Jonathan Bing use that scale to open their examination of advertising’s basic unit of value: the impression.

In Human Dot AI: How the Fight For Your Attention Became the World’s Biggest Business, the authors argue that automated advertising can create inventory faster than people can supply attention. The excerpt traces that imbalance from mass cable cancellations in 2016 to AI-generated campaigns.

Marchese and Bing write that, in 2016, people canceled cable subscriptions by the millions, ratings slipped and advertisers poured billions of dollars into digital platforms promising bigger audiences at much lower cost. Their account begins with a test of whether those platforms could reproduce television’s economics without delivering television’s attention.

What did three-second views do to the ad market?

In April 2016, Buzzfeed broadcast a live stream of two staffers in white coats and safety goggles wrapping hundreds of rubber bands around a watermelon. The watermelon eventually burst, and the stream drew more than 3 million views, according to Marchese and Bing.

CNBC called the broadcast “the future of television,” Marchese and Bing write. The authors note that no one could determine how much of the audience watched beyond a few seconds, including whether autoplay carried the video as users scrolled.

Every three-second view counted as an impression, which Marchese and Bing call advertising’s basic unit of value. They define an impression as an ad, or the content around it, appearing on a screen for a few seconds, whether anyone noticed it or not.

Marchese and Bing say advertising has always depended on attention, a finite resource because people have only 24 hours in a day. The algorithms behind digital advertising instead had one incentive, maximizing ad revenue, and learned to produce and promote whatever generated the most impressions.

When those algorithms ran short of real attention, they intercepted demand that already existed and claimed credit for purchases people would have made anyway, the authors write. That strategy set off a snowball effect: ad inventory surged, cheap and low-quality impressions flooded the market, and prices fell for every seller.

Marchese and Bing describe the outcome as a subprime market for human attention. They note that digital pricing erased distinctions buyers had accepted on television, where commercials carried different grades. Scrolling past an exploding watermelon became worth the same unit as watching every minute of a prime-time drama or a World Series game.

Why did premium television formats struggle?

Marchese and Bing say networks competing on digital’s terms ran more advertising minutes, which worsened the viewing experience and probably pushed more viewers toward streaming and ad-free platforms. “The less attention was worth, the more ‘impressions’ the industry had to manufacture to make up the difference,” they write.

As traditional television ads lost value, some networks cut ad loads and reconsidered the commercial break. Jonathan Bing, then in charge of advertising at Fox Networks Group, said he and his team tested six-second spots, premium slots at the start and end of breaks, and single-sponsor streams with minimal interruptions.

Marchese and Bing write that other networks pursued similar versions, promising fewer, better ads and offering advertisers more engaged audiences. The formats improved the exchange for viewers’ attention, but buyers resisted, asking why six seconds should cost more than 30.

The authors contrast those buyers with real estate and retail, where placement and scarcity carried value. In the ad market, brands wanted volume; meaningful change required advertisers to let agencies take risks, and most were unwilling to do so.

What does AI change now?

A decade later, Marchese and Bing say AI can generate an unlimited supply of content, filling feeds with AI-generated videos. The authors also note that platforms now automate every step of advertising, from writing an ad and targeting its audience to placing the buy.

Ads now appear inside AI assistants, according to Marchese and Bing. The placement puts the Reuters forecast of $1 billion in ChatGPT annual ad revenue inside an automated market capable of expanding impressions without a matching increase in human attention.

Marchese and Bing warn that endless content chasing finite attention will make every impression cheaper and less meaningful. They say the subprime market for attention will keep growing until advertising looks less like a market for attention and more like a toll on global commerce.

AI will keep optimizing what can be measured, the authors write, while culture, storytelling, trust and human connection remain hardest to measure and therefore easiest to neglect. The authors locate the deeper risk in that mismatch, not in any shortage of content.

Can AI reset the attention economy?

The same technology creates a reset opportunity, Marchese and Bing argue. AI can serve people rather than keep them scrolling, helping them manage time, filter information, make better decisions and reclaim attention.

Going forward, the authors argue that AI platforms should optimize for whether people recover attention, not merely whether impressions rise. Their proposed measure is whether AI helps users manage time, filter information and make better decisions, while leaving room for stories, experiences and genuine connection.