The new economics of cultural abundance

Culture is entering an experiment in abundance. Generative AI is reducing the time, labor and capital required to produce books, songs and images, allowing cultural goods to be created at a scale economically implausible only a few years ago. In publishing alone, monthly e-book releases on Amazon roughly tripled between late 2022 and late 2025, according to research by economists Imke Reimers and Joel Waldfogel, with AI accounting for essentially all of the increase.1

For Waldfogel, a professor at the University of Minnesota’s Carlson School of Management, AI belongs to a longer history of falling barriers to cultural production.

“What is happening in AI is an extension of a transformation that has been underway for more than two decades,” he told The Beiruter.

 Digitization dramatically lowered the cost of producing and distributing cultural products. AI is taking that process further.

But AI is pushing that process toward a more radical economic proposition, one in which supply can expand almost without limit while human attention cannot. In markets suddenly crowded with inexpensive content, value may instead accrue to authorship, and to the knowledge that a person invested time, judgment or emotion in making something.

The first collapse in costs

Generative AI did not begin the erosion of barriers to cultural production. Digitization did.

Waldfogel traces the first major change to the early 2000s, when falling production and distribution costs allowed works denied financing by traditional publishers, record labels, and studios to reach consumers directly.

Some became major commercial successes. Fifty Shades of Grey, The Martian, and Still Alice all began as self-published works before being acquired by traditional publishers.

“The period produced an explosion of successful products that wouldn’t have happened before,” Waldfogel said. 

There was a democratization made possible by lower-cost entry and a reduction in the barriers to entry into cultural production.

Generative AI carries similar economic logic into a different stage. Digitization had already made distribution cheap, while AI is now reducing the cost of production itself. The economic change is therefore not merely another expansion in access to markets but a reduction in the amount of human labor required to create what enters them.

More production, less appeal

Abundance, however, is not the same thing as demand.

In their January 2026 NBER working paper, Waldfogel and Reimers found that the surge in AI-containing Amazon e-books was accompanied by worse average sales ranks, numbers of ratings and star ratings than non-AI titles.

“With this giant influx of AI material, very little of it ends up near the top of the sales charts,” Waldfogel said.

Unlike the earlier phase of digitization, which produced successful works across the quality spectrum, the cost reduction from AI has so far mostly produced material with relatively little consumer appeal.

A similar gap between production and demand is emerging in music. Deezer, the French music-streaming service, said in July 2026 that it was receiving about 90,000 fully AI-generated tracks a day. At their June peak, they accounted for more than half of all new music delivered to the platform. Yet fully AI-generated tracks represented only 1% to 3% of listening.

The figures capture an emerging feature of AI cultural markets: supply can expand far faster than consumption. As the volume of available material rises, competition shifts from the ability to produce something toward the considerably harder task of persuading someone to consume it.

The authenticity premium

The distinction between human and machine production becomes more complicated when consumers cannot reliably perceive it themselves.

A November 2025 survey commissioned by Deezer asked 9,000 people in eight countries to distinguish fully AI-generated music from human-made tracks. Ninety-seven percent failed to identify the difference in a blind test. But 80% said fully AI-generated music should be clearly labeled, while 73% of streaming users wanted disclosure when platforms recommended it.

Research by Colleen Kirk, a professor of marketing at New York Institute of Technology, suggests this preference is tied less to technical quality than to what consumers believe communication represents.

Kirk and Julian Givi found consumers more accepting of AI for factual communications than emotional ones. When a message purported to express emotion, AI authorship reduced perceptions of authenticity because machines are not understood to possess the feelings being expressed.

“People value authenticity in human communications, especially when they are writing or communicating emotional content,” Kirk told The Beiruter.

Declaring that content, products, and other output are human-generated will likely become more important and valued by consumers.

The value of creative labour

For creators, generative AI presents a less straightforward economic trade-off. By lowering the labor required to produce creative work, it can make existing workers more productive while also creating new competition for what they produce. That competition can hurt creators even as it lowers costs and expands choice for consumers.

“Consumers win when producers face competition,” Waldfogel said.

A harder problem emerges if AI-generated work becomes good enough to divert substantial demand from human creators, weakening the incentives for people to continue producing the material on which cultural industries, and AI systems themselves, depend.

There is an irony embedded in that competition. Generative models capable of substituting for some creative labor were themselves built on enormous quantities of human-produced material and will require new material as they improve.

“Clearly, the AI companies need more training data, and so, in some sense, they need people to keep creating.” Waldfogel said.

Ultimately, there will need to be ways of ensuring that people continue to create, if for no other reason than to keep training the models.

AI may therefore produce an unexpected reversal. As the cost of making cultural products falls, the fact of human authorship can itself become part of what consumers value. The cheaper creation becomes, the more valuable it may be to know that someone actually created it.