Why the Myth of AI Persists: Lessons from Jaron Lanier’s 202
Key takeaways
- The term “AI” is a convenient but misleading shorthand for large statistical models and data pipelines.
- Anthropomorphizing machine‑learning systems creates a myth of agency that obscures corporate incentives and responsibility.
- Hype around AI fuels investment and policy avoidance, while diverting attention from concrete issues like bias, labor displacement, and accountability.
- Renaming tools, increasing transparency, and targeting human decision‑makers in regulations can mitigate the negative effects of the AI myth.
- Lanier’s critique urges a shift from sensational narratives to a grounded discussion about data, incentives, and the people who control them.
Published: July 2026 Inspired by Jaron Lanier’s New Yorker piece “there is no AI”
---
Introduction
When you scroll through today’s headlines, you’ll see phrases like “AI breakthrough,” “AI‑powered,” or “AI revolution.” Yet, as Jaron Lanier reminds us, the term artificial intelligence is more a cultural shorthand than a scientific description. In his 2023 essay, Lanier contends that there is no singular entity called “AI” – only a collection of statistical models, data pipelines, and corporate strategies that we have collectively labeled as such. This post explores his argument, why the myth endures, and what a more precise vocabulary could mean for developers, policymakers, and the public.
---
The Semantics of “AI”
Lanier’s first point is linguistic: intelligence is a property of organisms that can learn, reason, and adapt in open‑ended ways. The systems we call AI—large language models, image generators, reinforcement‑learning agents—are fundamentally pattern‑matching machines trained on massive datasets. They do not possess goals, self‑awareness, or any form of consciousness.
By anthropomorphizing these tools, we grant them a status that obscures the true source of their behavior: the data fed to them and the incentives of the companies that deploy them. When a model like ChatGPT produces a witty paragraph, we tend to attribute the cleverness to the model itself, rather than to the engineers who curated the training corpus, the heuristics that guide token selection, and the profit motives that shape its fine‑tuning.
The Illusion of Agency
Lanier argues that the illusion of agency is not accidental. It serves several functions:
1. Marketing – Claiming that a product is “AI‑driven” adds a veneer of futurism and justifies premium pricing. 2. Legal Shielding – If a system is framed as an autonomous “intelligence,” responsibility can be deflected onto the technology rather than the creators. 3. Cultural Narrative – The story of humanity confronting a new form of intelligence taps into deep myths about creation, hubris, and apocalypse, making the technology more news‑worthy.
The result is a feedback loop: hype fuels investment, which fuels more hype, and the underlying technical reality remains opaque to most users.
Economic and Ethical Stakes
Lanier’s critique is not merely semantic; it has concrete implications:
- Labor Displacement – By portraying automation as an inevitable march of “intelligent” machines, we sidestep the policy discussion about how to redistribute the gains from increased productivity. - Bias and Power – When a model’s output is blamed on “AI,” we ignore the fact that the training data reflect existing societal biases, and that the companies controlling the data pipelines wield disproportionate influence over public discourse. - Accountability – If a model generates harmful content, who is liable? The answer becomes murkier when the technology is personified as an autonomous actor.
Lanier urges us to re‑center the conversation on people, data, and incentives, rather than on an imagined digital mind.
Reframing the Conversation
How can we adopt Lanier’s perspective without discarding useful shorthand? Here are three practical steps:
1. Rename the Tools – Use terms like large‑scale statistical model or data‑driven language system when discussing capabilities. Reserve “AI” for research that genuinely tackles open‑ended problem solving (e.g., symbolic reasoning, robotics with embodied cognition). 2. Make the Pipeline Visible – Publish data provenance, model architecture choices, and fine‑tuning objectives. Transparency turns the “black box” into a process that can be audited. 3. Shift Responsibility – Frame ethical guidelines around developers and deployers rather than the technology itself. Regulatory frameworks should target the human decision‑makers who set objectives, collect data, and monetize outcomes.
By reshaping the language, we can reduce the mystique that lets powerful actors evade scrutiny.
---
Conclusion
Jaron Lanier’s 2023 essay is a reminder that language shapes perception. The term artificial intelligence has become a cultural catch‑all that masks the real drivers of today’s machine‑learning boom: massive datasets, compute power, and profit motives. Recognizing that there is “no AI” in the literal sense does not diminish the transformative impact of these technologies; it simply calls for a more honest dialogue about who builds them, how they work, and who benefits.
If we adopt Lanier’s call for precision, we can move beyond sensationalism toward policies and practices that hold the right actors accountable and ensure that the benefits of advanced statistical models are distributed fairly. The future of technology depends not on the myth of an autonomous intelligence, but on our collective ability to govern the very systems we create.
---
Feel free to share your thoughts in the comments. How do you think we can best re‑educate the public about the true nature of these systems?
Sources: https://www.newyorker.com/science/annals-of-artificial-intelligence/there-is-no-ai