“There’s nothing “artificial” about AI.”
Fei-Fei Li offered this compact correction in written testimony to the U.S. House of Representatives in June 2018. “There’s nothing ‘artificial’ about AI,” she wrote, before stressing that the technology is inspired by people, created by people, and consequential for people. The statement challenged a convenient metaphor. Artificial intelligence can sound like a force that arrived from outside society, complete with its own momentum and values. Li instead located it within human history: researchers decide what problems to solve, organizations decide where systems are deployed, and communities live with the benefits, mistakes, and unequal effects that follow.
Li’s own work illustrates why that framing matters. Her leadership of ImageNet helped accelerate progress in computer vision by organizing a vast labeled data set and a benchmark around it. The breakthrough was technical, but it also depended on human categorization, labor, and judgment. Computer vision systems do not simply discover a neutral world. They learn from selected examples and definitions, then operate in settings where social context can differ sharply from a laboratory test. Calling AI human-centered means treating those inputs and environments as part of the engineering problem, rather than as background details to address after a model is built.
The quote also reframes common debates about whether AI will be good or bad. Technology alone does not settle that question. A model that improves medical imaging, expands access for people with disabilities, or helps scientists analyze complex data can serve public goals. The same underlying capabilities can also enable intrusive surveillance, automate unfair decisions, or concentrate power. Outcomes depend on whose needs guide design, who is represented in data, what safeguards govern deployment, and whether affected people have a meaningful way to challenge errors. Human values enter the system long before the first user encounters its output.
For leaders, Li’s sentence is a demand for responsibility rather than fear or hype. If AI is a human creation, accountability cannot be delegated to an algorithm. Teams need diverse expertise, careful measurement, transparent limits, and feedback from the people most likely to bear the risk. Policymakers must protect innovation without pretending that markets automatically resolve every harm. Users, meanwhile, deserve enough information to judge when a system is assisting them and when it is making consequential choices. Public participation belongs in that process from the beginning. The future of AI is not something machines simply deliver; it is something people build, govern, and continuously renegotiate together.
Li submitted the statement for a 2018 House hearing on artificial intelligence and the future of innovation. She paired the line with a direct reminder that AI is inspired, created, and used by people.
Her testimony urged the United States to invest in long-term research, education, and a workforce able to develop the technology while understanding its social consequences.
Human-centered design begins with the people affected by a system, not only the performance of its model. It asks whether the problem is worth solving and whether the result works fairly in real settings.
It also widens the technical team. Social scientists, domain experts, community members, and policymakers can reveal risks or needs that a narrow benchmark will not capture.
Today’s generative systems make AI feel increasingly autonomous, yet their data, objectives, interfaces, and deployment rules remain the products of human decisions. Li’s wording keeps those decisions visible.
The practical lesson is accountability: organizations cannot blame “the algorithm” for outcomes they chose to automate. People retain the duty to test, explain, correct, and govern the systems they release.
Explore more "Quotes of The Day"
Continue exploring how writers, thinkers, artists, and advocates have described the choices that shape an everyday life.