Generative AI is on track to become an entertainment-first industry, driven by commercial incentives and youth adoption, yet economic evaluation and policy frameworks are unprepared to measure its cultural and welfare effects. The authors propose a 'thick entertainment' lens to treat AI-generated entertainment as socially consequential and call for new metrics, empirical studies, and regulatory attention beyond harm mitigation.

AI as Entertainment

Cody Kommers, Ari Holtzman · Fetched March 15, 2026 · arXiv.org

semantic_scholar theoretical low evidence 7/10 relevance Full text usable extracted full text DOI Source PDF

The paper argues generative AI is likely to evolve into an entertainment-first industry, creating cultural and welfare effects that current productivity-focused evaluation frameworks and governance tools fail to capture.

Generative AI systems are predominantly designed, evaluated, and marketed as intelligent systems which will benefit society by augmenting or automating human cognitive labor, promising to increase personal, corporate, and macroeconomic productivity. But this mainstream narrative about what AI is and what it can do is in tension with another emerging use case: entertainment. We argue that the field of AI is unprepared to measure or respond to how the proliferation of entertaining AI-generated content will impact society. Emerging data suggest AI is already widely adopted for entertainment purposes -- especially by young people -- and represents a large potential source of revenue. We contend that entertainment will become a primary business model for major AI corporations seeking returns on massive infrastructure investments; this will exert a powerful influence on the technology these companies produce in the coming years. Examining current evaluation practices, we identify a critical asymmetry: while AI assessments rigorously measure both benefits and harms of intelligence, they focus almost exclusively on cultural harms. We lack frameworks for articulating how cultural outputs might be actively beneficial. Drawing on insights from the humanities, we propose"thick entertainment"as a framework for evaluating AI-generated cultural content -- one that considers entertainment's role in meaning-making, identity formation, and social connection rather than simply minimizing harm. While AI is often touted for its potential to revolutionize productivity, in the long run we may find that AI turns out to be as much about"intelligence"as social media is about social connection.

Summary

Main Finding

Generative AI is rapidly becoming an entertainment technology as much as (or more than) an intelligence-for-productivity technology. This shift is already visible in user behavior (especially among youth), platform design, and corporate strategy. Because entertainment markets are large and monetizable, they will strongly shape AI product development, yet the AI field lacks evaluation frameworks and economic tools to measure and steer AI’s cultural and welfare impacts beyond harm mitigation. The authors propose “thick entertainment” — an evaluative lens that captures entertainment’s roles in meaning‑making, identity, and social connection — as a way to assess positive cultural value, not only risks.

Key Points

Observed behavior and platforms
- Large-scale adoption of AI for entertainment: Character.AI (estimated 20–28M monthly active users; ~53% under 25), Vtubers like Neuro-sama (millions of views, large paying subscriber bases), AI-first short-video platforms (e.g., Sora).
- AI-generated music and podcasts entering mainstream consumption; AI artists charting on Billboard; Google’s Notebook LM offering AI-generated podcasts.
- Children and teens are heavy users: US survey of 1,060 teens (13–17) found 73% had used AI companions; UK survey (ages 9–17) found 25% used chatbots “just for fun,” and 58% of 9–12-year-olds reported chatbot use.
Motivation and usage dynamics
- Studies indicate motivation to use AI for entertainment grows over time and can exceed instrumental (productivity) motivation.
- Entertainment use tolerates more creativity/hallucination; gaps that hurt productivity apps may be acceptable or even desirable in entertainment contexts.
Business model and industry forces
- Massive capital and infrastructure commitments (e.g., multi‑GW compute deals, hundreds of billions invested across industry) create pressure to monetize at scale.
- Entertainment and media industries constitute sufficiently large markets (estimates: music streaming ~$20B, short-form video ~$190B, games ~$450B, projected global media & entertainment ~$3.5T by 2029) to be a primary monetization path.
- Big AI firms are actively pursuing entertainment plays (e.g., OpenAI licensing Disney IP for Sora; Google and Meta releasing entertainment‑oriented products), and legacy media companies are integrating AI into content pipelines (BBC, Financial Times).
Conceptual and evaluative gap
- AI research and benchmarking prioritize intelligence, and assessments emphasize safety and harms; there is little shared language or metrics for positive cultural effects.
- The authors argue for “thick entertainment” — evaluating AI outputs on how they contribute to meaning, identity, empathy, social connection, and other non-instrumental cultural values.
Social risks and externalities highlighted
- Attention diversion, changes to meaning-making and identity formation (especially among youth), increased parasocial relationships, shifts in content incentives, cultural concentration, IP/licensing dynamics, and labor impacts in creative industries.

Data & Methods

Approach: conceptual paper combining literature review, industry reporting, platform case studies, and synthesis of survey and usage data to build an argument about directional change and its implications.
Empirical inputs cited include:
- Platform usage statistics and interaction counts (e.g., Character.AI public reports and third‑party estimates).
- Case examples of AI entertainment (Neuro-sama, Sora, Vtubers, AI music on streaming platforms).
- Surveys:
  - US teen survey (n = 1,060; Spring 2025): 73% used AI companions.
  - UK survey of children and teens (n = 1,000; ages 9–17): 25% used chatbots for fun; many under‑13 reported usage.
  - Longitudinal/experimental user study on motivations (5-week usage study measuring entertainment vs. instrumental motivation).
- Industry and market reports: AISI frontier AI trends, McKinsey State of AI, news reporting on compute deals (NYT), market size estimates for media sectors.
Methodological limits acknowledged by the authors:
- Evidence is early-stage and often relies on secondary sources, platform disclosures, industry reports, and surveys with potential sampling limitations.
- No large-scale causal or longitudinal econometric analysis yet; the paper is intended to motivate further empirical research and framework development.

Implications for AI Economics

Revenue and R&D allocation
- Expect significant reallocation of R&D and product design toward entertainment use cases because entertainment markets can justify massive infrastructure costs; economists should model firm incentives under large fixed costs and rich entertainment demand.
- Business models to watch: subscriptions, micropayments, in-app economies, IP licensing, creator monetization, and platform fees tied to parasocial engagement.
Market structure and competition
- Entertainment-driven returns can entrench large incumbents (who control compute, data, and IP deals), raising concentration and winner‑take‑most dynamics; antitrust and industrial organization analyses should assess barriers to entry and counterfactual innovation paths.
Valuation of cultural goods and welfare
- Standard productivity measures (e.g., GDP, labor productivity) understate welfare effects from entertainment consumption. New metrics are needed to value cultural goods, including non-market benefits (meaning, social capital) and harms (attention externalities, mental health).
- “Thick entertainment” implies multidimensional welfare measurement — economists should develop tools to quantify identity formation, social cohesion/dissolution, and cultural capital, possibly via structural models, revealed preference in attention markets, or stated-preference surveys.
Attention economy and externalities
- Economists should model attention allocation between productivity and entertainment AI, including substitution/complementarity, externalities from content personalization, and social welfare impacts of increased attention capture.
- Policy implications include consideration of platform regulation, content taxation/subsidy, disclosure/labeling rules, and interventions targeted at youth.
Labor, IP, and distributional effects
- Creative labor markets may face displacement, re-skilling pressures, and altered bargaining power; study wage effects, task reallocation within creative sectors, and the role of AI in augmenting vs. replacing creators.
- IP and licensing (e.g., Disney deals) will shape access and rents; economists should model licensing market power and downstream effects on variety, quality, and consumer surplus.
Research priorities (recommended)
- Empirical measurement: longitudinal user-level data on time use, spending, learning, and wellbeing tied to AI entertainment consumption.
- Demand estimation: elasticity of entertainment demand to quality, novelty, personalization, and price; cross-elasticities with productivity AI.
- Welfare accounting: methods to incorporate cultural benefits and harms into cost‑benefit analyses of AI deployment.
- Market design: optimal platform incentives, creator revenue-sharing schemes, and policies to mitigate attention externalities while preserving cultural innovation.
- Youth-focused effects: targeted studies on development, identity formation, and regulatory approaches to protect minors.

In short: Generative AI’s likely monetization via entertainment markets will shape product features, incentives, and societal effects. AI economists need new models, data collection, and welfare metrics (including the proposed “thick entertainment” dimensions) to evaluate and guide this transition.

Assessment

Paper Typetheoretical Evidence Strengthlow — The paper is primarily conceptual and argumentative, relying on selective early usage statistics, surveys, industry trends, and analogies to attention-economy platforms rather than systematic, representative, or causal empirical analysis; it proposes hypotheses and a framing rather than providing causal estimates or robust quantitative validation. Methods Rigormedium — The paper uses interdisciplinary literature synthesis and reasoned argumentation grounded in media studies, cultural theory, and observable industry signals, which is appropriate for a conceptual contribution; however, it lacks pre-registered empirical designs, representative datasets, quasi-experimental variation, or robustness checks that would raise methodological rigor to a high level. SampleNo original large-scale dataset; evidence consists of synthesized interdisciplinary literature, selective early usage statistics and surveys (not necessarily representative), industry revenue and business-model trend data, and illustrative examples from attention-economy platforms and creator ecosystems. Themesadoption productivity org_design governance GeneralizabilityConceptual claims are not empirically validated and may not hold across countries with different cultural industries or regulatory regimes, Early-adopter patterns (notably youth usage) may not generalize to older or enterprise users, Business-model trajectories depend on firm strategies, regulation, and technological advances that could change rapidly, Cultural impacts and 'thick' meanings are context-dependent and may vary substantially across languages, subcultures, and media formats

Claims (11)

Claim	Direction	Outcome	Confidence & Evidence	Details
Generative AI systems are predominantly designed, evaluated, and marketed as intelligent systems which will benefit society by augmenting or automating human cognitive labor, promising to increase personal, corporate, and macroeconomic productivity. Firm Productivity	positive	promised productivity gains (personal, corporate, macroeconomic) / positioning of AI systems	Reading fidelity medium Study strength low	not reported 0.04
This mainstream narrative about what AI is and what it can do is in tension with another emerging use case: entertainment. Adoption Rate	mixed	dominant narrative versus emerging use-case prevalence (productivity-oriented vs entertainment-oriented)	Reading fidelity medium Study strength low	not reported 0.04
The field of AI is unprepared to measure or respond to how the proliferation of entertaining AI-generated content will impact society. Ai Safety And Ethics	negative	readiness/preparedness of AI research and evaluation frameworks to assess societal impacts of entertaining AI content	Reading fidelity medium Study strength low	not reported 0.04
Emerging data suggest AI is already widely adopted for entertainment purposes — especially by young people — and represents a large potential source of revenue. Adoption Rate	positive	adoption rates for entertainment use (particularly among young people) and potential revenue from entertainment use	Reading fidelity medium Study strength low	not reported 0.04
Entertainment will become a primary business model for major AI corporations seeking returns on massive infrastructure investments. Firm Revenue	positive	share of corporate business models/revenue derived from entertainment for major AI firms	Reading fidelity low Study strength low	not reported 0.02
This (entertainment-as-business-model) will exert a powerful influence on the technology these companies produce in the coming years. Innovation Output	mixed	product design priorities and technological development directions influenced by entertainment business incentives	Reading fidelity low Study strength low	not reported 0.02
Current AI evaluation practices show a critical asymmetry: while AI assessments rigorously measure both benefits and harms of intelligence, they focus almost exclusively on cultural harms. Ai Safety And Ethics	negative	scope and balance of AI assessment metrics (coverage of benefits vs cultural harms)	Reading fidelity medium Study strength low	not reported 0.04
We lack frameworks for articulating how cultural outputs might be actively beneficial. Ai Safety And Ethics	negative	existence/availability of evaluative frameworks that characterize positive cultural impacts of AI-generated content	Reading fidelity medium Study strength low	not reported 0.04
The authors propose 'thick entertainment' as a framework for evaluating AI-generated cultural content — one that considers entertainment's role in meaning-making, identity formation, and social connection rather than simply minimizing harm. Other	positive	presence and scope of the proposed evaluation framework ('thick entertainment') for cultural content	Reading fidelity high Study strength low	not reported 0.06
AI is often touted for its potential to revolutionize productivity. Firm Productivity	positive	prevalence of claims asserting AI-driven productivity improvements	Reading fidelity high Study strength low	not reported 0.06
In the long run we may find that AI turns out to be as much about 'intelligence' as social media is about social connection (i.e., AI may be primarily about entertainment/social connection rather than productivity). Consumer Welfare	mixed	relative cultural role of AI (entertainment/social connection) compared to productivity orientation over the long term	Reading fidelity low Study strength low	not reported 0.02