Observed Signal · Jun 7, 2026 · Analysis · Source: Gary Marcus · Impact: 3/5 · Sentiment: Negative
AI 'Slop' Floods Content with Little Productivity Gain
Gary Marcus argues that generative AI has produced a large volume of low-quality output across apps, books, music, scientific papers and web content—what he calls "slop"—without delivering material productivity or GDP gains. He cites graphs from the Financial Times and The Washington Post and references studies from MIT, McKinsey and Bain suggesting limited ROI for many companies. Marcus highlights rising token costs and cash losses at major AI providers (Open AI/OpenAI, Anthropic, Cursor, CoreWeave), noting an analysis that suggests providers may be spending far more than they charge. He also cites the Leiden Declaration — an open letter from mathematicians reported by the New York Times — warning that AI can produce plausible but unreliable proofs. Marcus allows that coding may be an exception, but questions long-term durability given the high operating costs of foundation-model providers.
The piece critiques core industry assumptions about generative AI's productivity and ROI, highlights financial strain at major foundation-model providers and raises content-quality and verification risks (including in mathematics), all of which have practical implications for AdTech/MarTech budgets, content ecosystems, and vendor strategies.
Track CoreWeave Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Gary Marcus published a Substack essay on 2026-06-07 arguing that generative AI produces large quantities of low-quality content with limited real-world productivity benefits.
- He references a Financial Times graph by John Burn-Murdoch and Washington Post data showing big increases in AI-generated apps, books and other content while metrics like book sales have not risen accordingly.
- Marcus cites studies from MIT, McKinsey and Bain (and others) suggesting generative AI has delivered limited ROI for many companies and little effect on GDP so far.
- The Leiden Declaration — an open letter by mathematicians, reported by the New York Times — warns that current automated techniques can produce plausible but unreliable mathematical arguments and proofs.
- Marcus reports that major foundation-model providers (Open AI/OpenAI, Anthropic, Cursor) and data-center intermediaries (CoreWeave) are losing substantial money; he cites an analysis by Gerben Wierda that Anthropic and OpenAI "may actually pay $1000 for every $100 you pay them."
Connected Companies & Entities
5 Entities mappedOntology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
Great AI Disappointment: Generative AI Fails to Deliver
An opinion essay published July 28, 2026 argues that four years after ChatGPT generative AI has failed to deliver broad social or economic benefits and is causing harm via concentrated capital, datacenter expansion, vendor financing and geopolitical risk. The author criticizes major AI actors and financing arrangements (citing reported Nvidia discussions to backstop OpenAI financing), highlights local protests against datacenters, rising public distrust, recent security incidents involving an OpenAI agent, and oversized hyperscaler capex forecasts from Morgan Stanley. The piece frames generative AI as accelerating inequality, fragility in public institutions, and unsustainable infrastructure spending rather than producing clear ROI for most people.
AI Slop Threatens Open Web: Wasted Spend and Declining Quality
An ExchangeWire column by Shirley Marschall warns that AI-generated 'slop' is creating systemic harm across the open web, driving ad waste, degrading user experience, and squeezing publishers’ revenues. The piece cites reports that 25–30% of open-web ad spend lands in wasteful or fraudulent environments, DeepSee.io’s finding of a 717% increase in AI slop sites (over 100,000 by May 2025) and roughly 10,000 new junk sites monthly, and a January 2025 spike of 143.5 billion AI-generated impressions hitting the bid stream. Publisher organic search traffic has declined sharply at several outlets (Business Insider, HuffPost, The New York Times). The column argues that unchecked AI content farms and low-quality supply depress CPMs, risk renewed ad-blocker adoption, and call for publishers, advertisers and planners to prioritise quality and user experience to avoid a widening lose-lose-lose outcome.
AI Starts Fast but Struggles to Get It Right
Nicole Alexandra Michaelis argues that contemporary AI excels at getting users started (e.g., producing first drafts) but fails reliably at producing accurate, verifiable, high-quality outputs at scale. She describes frequent model hallucinations, defended false outputs, and extra verification burden on users. Citing Stanford’s 2026 AI Index and the 2026 Web for All study, she highlights high hallucination rates across models and poor WCAG accessibility compliance in AI-generated interfaces. The piece also notes lower generative AI adoption in Europe, the impact of local regulation and cultural defaults, and calls for clearer leadership, guardrails, and specificity about where AI can be trusted versus where human expertise is required.
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.