Kevin Kelly on AI, New Platforms and a Life of Learning

Zhang Xiaojun introduces Kevin Kelly, the founding executive editor of Wired and author of Out of Control, What Technology Wants and The Inevitable. The interview spans AI, platforms, borders and life advice.

An interviewer identifying herself as a Tencent News reporter begins by asking about his long beard. Kelly says it began from a reluctance to shave, not a carefully chosen symbol.

Asked about ChatGPT, Kelly says the existence of such a system did not surprise him, but the use of large language models to achieve it did. This discussion reflects his views at the time of the interview.

He cautions that useful AI deployment may take longer than people expect. Replacing a task involves changing organizations and infrastructure as well as building a model, he argues.

Kelly expects an AI-related company eventually to become a major platform, but says it is too early to know whether that will be OpenAI, Nvidia or another firm. This is a prediction, not a statement that any firm had displaced its competitors.

His historical theory is that dominant technology companies are often challenged from outside their original category: he uses IBM, Microsoft, Google and Facebook as examples of different waves.

He says being first is not sufficient: later entrants may execute a product better. In AI, he argues, the field was still early enough for other companies to develop different approaches.

Asked about Sam Altman, Kelly offers a personal impression of his temperament and compares him with other technology executives. He presents this as opinion, not a prediction of a specific company outcome.

Kelly calls the “mirror world” a digital layer over the physical world. He thinks it would require lightweight, durable smart glasses and systems that can understand what they see.

He compares the available devices of that period to the pre-smartphone era: useful steps, but not yet the interface needed for widespread adoption.

In his envisioned third platform, AI would identify objects and relationships in a room, while wearable devices could scan the environment as people use them. He presents this as a possible future architecture.

He speculates that a startup could develop the next major platform because it can serve a small market and experiment without the scale expected of a large incumbent. He also expects competitors to follow if a product works.

Asked about US and Chinese technology giants, Kelly says they are unlikely in his view to dominate every future wave. He distinguishes losing dominance in a new category from disappearing as a company.

On borders, he argues that technology and knowledge move across countries even when governments impose restrictions. He favors open trade and considers the long-term effects of restrictions harmful; these are his policy opinions.

Asked for advice to Chinese entrepreneurs facing cross-border difficulties, he explicitly says he does not know enough about their financing conditions to give a confident answer.

He describes Asian hardware and US software ecosystems as relative strengths at the time, emphasizing networks of suppliers and supporting organizations. He adds that the balance could change.

He predicts that a Chinese electric vehicle could become a globally desired consumer brand, while acknowledging he had not examined the particular companies raised in the question.

Kelly sees both convergence and difference across cultures: offices and clothing may look similar, while relationships, work practices and information flow can remain distinct.

On AI safety, he considers a scenario in which AI treats humans as pets unlikely. His argument is that intelligence alone does not automatically give an actor control over the physical world; this is his assessment, not a settled safety finding.

He imagines many specialized AIs rather than one uniform system. In his view, their value could include creating new choices and forms of work as well as assisting existing tasks.

Discussing growth, he says he did not see exponential growth in intelligence at the time, despite rapid growth in computing resources. He is making an empirical claim and forecast from that period, not stating a universal law.

He describes AI as an enabling technology that may speed work in other fields, citing biological research and drug discovery. His description of protein folding is broad; the interview does not establish that every related scientific problem has been solved.

There is no single AI business model, he argues, because there are many systems and applications. Hardware and infrastructure can matter alongside software, and economic benefits may take time to appear.

Asked whether AI will become an elite-only field, he says ordinary users can learn by exploring tools and developing practical expertise. His broader point is that new fields can create opportunities outside formal credentials.

On returning to China, Kelly says he had visited many times but had not yet seen enough during this visit to assess recent changes. He also offers a personal observation about concern over political polarization in the United States.

His stated life goal is to remain a lifelong learner, change his mind when warranted and help people gain tools and options for becoming their best selves.

He encourages working in emerging areas that may not yet have a familiar name. When Wired began, he says, its focus on the culture of technology did not fit an obvious magazine category.

Out of Control took him five years to write. He says he wrote the book he wanted to read and used writing to discover and develop his own thinking about biology and machines.

He became interested in technology when the internet made it feel more human-scale and connected to everyday life. He sees conversational AI as another example of technology at that scale.

His high-school role model was Henry David Thoreau, whose simple living and close observation of nature appealed to him. Kelly later built his own home.

Reflecting on fame and resources, Kelly says he tries to stay close to simple experiences through travel and camping. He worries that abundant money can make it easy to buy solutions instead of inventing them.

He provocatively advises against acquiring a billion dollars because managing great wealth can consume time and attention. This is his personal philosophy, not financial advice or a claim that wealth affects everyone identically.

For someone who is already successful, he recommends remaining a beginner in an unfamiliar area where failure is possible. He believes this can help preserve curiosity and creativity.

Asked how much money is enough, he says the answer depends on being able to do what matters to oneself. He contrasts this with automatically assuming that twice one’s current wealth is the right amount.

He says his family supports nonprofit work, including the Long Now Foundation, and also intends to provide for their children.

On aging, he says physical abilities change, but he tries to stay mentally young by remaining willing to change his mind. He values the perspective accumulated through many cycles and friendships.

His current project imagines a desirable world a hundred years ahead with AI, genetic engineering and monitoring. He wants to ask what kind of future people would actually want to inhabit.

A question that continues to interest him is whether evolution has a direction and, if so, how that direction arises. He says he does not know the answer.

Asked what children should study for future jobs, he says he cannot identify specific future roles. He recommends developing the general ability to learn quickly, recalling that Wired hired for aptitude when the web was new.

Asked about competitive schooling, he argues that children need unscheduled time to play rather than filling every moment with study. This is his view on education.

On his “1,000 True Fans” idea, he sees AI as a possible assistant for administrative work and for individual creators making media. He says he would be cautious about using AI as a substitute for direct relationships with fans.

If AI interacts with people, Kelly believes the interaction should be disclosed. He says users may accept AI assistance when they know what they are interacting with.

Zhang closes the program and invites listeners to follow the podcast.