“As a computer scientist, I don’t like the term ‘A.I.’”
Computer scientist and virtual-reality pioneer Jaron Lanier opened a 2023 essay with this objection. He does not deny the impressive capabilities of modern models. He questions a term that encourages people to imagine an independent, almost mythical intelligence arriving from outside society. Large models are built from human-created data, engineered objectives, vast computing systems, and repeated decisions about deployment. Calling the result “A.I.” can make those contributors disappear and can allow companies to present choices as the inevitable actions of a machine.
Language changes how responsibility is assigned. If a chatbot “decides” to produce harmful content, attention may drift away from the organization that selected its training process, interface, safety measures, and release conditions. If an automated tool “creates” an image, the writers, artists, photographers, and communities represented in its training data may be treated as raw material rather than participants in a value chain. Personifying software can therefore grant the machine credit while granting its owners power and leaving affected people with little recognition or recourse.
Lanier proposes a more human-centered way to understand computation. Models recombine patterns made available through collective cultural production. Their outputs can be useful and surprising, but the surprise need not establish a new person. This framing opens economic questions. Could systems track meaningful contributions, compensate sources, or create new forms of collective bargaining around data? Could interfaces help users recognize provenance and uncertainty? These proposals are difficult, yet they begin from the premise that technology should strengthen relationships among people rather than replace those relationships with a fictional machine subject.
The debate is not merely about vocabulary. Metaphors guide investment, law, and everyday behavior. A user who believes a system understands may reveal more than intended or trust advice beyond its competence. A manager who believes intelligence has become autonomous may undervalue employees whose knowledge made the workflow possible. More precise descriptions—language model, recommender, classifier, prediction system—can clarify what a tool does and where it fails. Lanier’s provocation asks society to celebrate computational achievement without surrendering human agency to the story told about it. Lanier’s objection is ultimately about agency. Once people can see the human sources and institutional decisions inside a system, they can argue about credit, compensation, consent, and responsibility instead of treating technology as an independent force. That clarity can support better products as well as fairer relationships between model builders, users, and the creators whose work supplies cultural knowledge.
The term AI can encourage people to imagine an independent mind instead of a system built from human data and decisions.
That metaphor influences trust, investment, labor policy, and where the public looks for accountability.
Models learn patterns from cultural work produced by writers, artists, photographers, programmers, and communities.
Treating the machine as the creator can obscure contribution, consent, compensation, and the owners who capture value.
Terms such as language model, classifier, recommender, or prediction tool identify a narrower function and failure pattern.
Precision helps users calibrate trust and keeps organizations responsible for how a system is designed and deployed.
Explore more "Quotes of The Day"
Continue exploring how writers, thinkers, artists, and advocates have described the choices that shape an everyday life.