Kevin Kelly's 2049: A Future Map for Chinese Entrepreneurs

On August 7, 2025, OpenAI released GPT-5.

Ethan Mollick, a professor at the Wharton School, has long been concerned about innovation and entrepreneurship as well as artificial intelligence. The next day, he shared his experience of using GPT-5 on his blog. He only input one sentence: "Create a Brutalist architecture generator, make it cool." Two minutes later, a 3D city simulator that could be freely dragged and edited appeared on his screen: there were neon lights, speeding cars, and a dynamic camera. He didn't look at a single line of code throughout the process. Even in the subsequent optimization, he just kept saying "Please make it better."

GPT-5 added a saving system, preset building types, and a facade editor on its own. It wasn't just executing commands; it was "doing things," deciding on its own what to do, how to do it, and to what extent.

This scenario reminds me of a core idea in Kevin Kelly (nicknamed K.K.)'s new book 2049: Possibilities in the Next 10,000 Days: We've always been trying to make AI think like humans, but the real strength of AI lies precisely in the fact that it doesn't think like humans.

K.K. calls this "artificial aliens," meaning they are more like aliens than human 2.0. Just as an airplane can fly without flapping wings, AI can "think" without imitating the human brain. GPT-5 proves this: No human would, while writing a passage, actually play seven word games simultaneously (increasing one word in each sentence, spelling out a secret message with the first letters of each sentence, and having alliteration within each sentence) and still keep the passage coherent.

One month before the release of GPT-5, NVIDIA's market value exceeded $4 trillion, becoming the first company in human history to reach this market value milestone. It only took NVIDIA 13 months to go from $3 trillion to $4 trillion. Some say it's a bubble, while others say it's just the beginning. But when you see the performance of GPT-5, you'll understand: This isn't just hyping up AI; it's that we haven't fully grasped what AI will bring.

As the founding editor-in-chief of Wired and the author of Out of Control and What Technology Wants, K.K. turned his attention to China 25 years later this time①. So, he found a Chinese collaborator, Wu Chen, the former editor-in-chief of The Economist Business Review. The two completed this "Sino-US co-authored" book through a dialogue.

After reading this book, my biggest takeaway is: While everyone is asking "Will AI replace humans?", K.K. is asking a more interesting question: "What should humans do when AI starts to 'do things' on its own?"

#1

Mirror World: The Next Trillion-Dollar Hypothesis

The most core concept K.K. proposes in the book is the "Mirror World," a world where reality and virtuality are completely integrated.

This isn't the metaverse that Mark Zuckerberg has poured hundreds of billions of dollars into and is still losing money on. The Mirror World K.K. talks about is more down-to-earth: When you walk on the street wearing smart glasses, you can see the real-time ratings of each restaurant; when you work in a warehouse, a pair of "shadow hands" will appear in your glasses, teaching you how to operate complex machines; medical students can "swim" in a virtual heart and disassemble each valve with their own hands.

Does it sound like science fiction? Actually, Apple's Vision Pro has already started trying it, but the unbearable weight on the neck and the $3,500 price have discouraged most people. However, K.K. believes that the key to the Mirror World isn't hardware, but AI.

He has a very insightful judgment: The main use of large-scale AI in the future will be to power the Mirror World. Why? Because overlaying a digital world on the real world requires processing an astronomical amount of data - everyone's expressions, movements, and the environment need to be captured, analyzed, and rendered in real-time. Without powerful and affordable AI, the Mirror World would be a joke.

This reminds me of the bike-sharing war in 2016. At that time, everyone was discussing business model innovation, but what really determined the winner was who could reduce the cost of smart locks from $300 to $30. The same goes for the Mirror World. The key isn't how cool the concept is, but when the cost of AI can drop to rock-bottom prices.

What's even more interesting is one of K.K.'s predictions: In the future, the company that dominates the Mirror World may not be from China or the United States, but an Indian company with one million employees globally. The reason is simple: As a relatively independent third party, India has both engineering capabilities and can play both sides off against the middle. Although this judgment is bold, it makes sense.

2049: Possibilities in the Next 10,000 Days Author: [US] Kevin Kelly (Author) / Wu Chen (Editor) Publisher: CITIC Publishing Group

#2

Artificial Aliens: Why AI Will Never Become Human

If the Mirror World is K.K.'s judgment on the future form, then "artificial aliens" is his understanding of the essence of AI. This term is very interesting. K.K. believes that we shouldn't understand AI as "artificial intelligence," but as "artificial aliens." Because their way of thinking is completely different from humans, just like aliens.

For example, the way AlphaGo plays Go is completely beyond the comprehension of human players. It isn't imitating humans; it's "thinking" in a completely different way. This is like an airplane achieving flight with fixed wings instead of imitating birds flapping their wings.

K.K. divides the capabilities of AI into two categories:

• Lowercase creativity (creativity): Replication, optimization, and efficiency improvement;

• Uppercase creativity (CREATIVITY): Breakthrough, innovation, and paradigm change.

In his view, in the next 25 years, AI will fully master lowercase creativity. For example, writing copy, drawing illustrations, and programming - these jobs will all be replaced by AI. But when it comes to uppercase creativity, such as proposing the theory of relativity, inventing the transistor, or creating One Hundred Years of Solitude, AI still has a long way to go.

What does this view mean for corporate managers? Don't expect AI to help you make strategic decisions, but you can let it help you with tactical execution.

I remember chatting with a manufacturing boss some time ago. He said that the accuracy of the AI quality inspection system in the factory in identifying defects has already exceeded that of humans. But when it comes to letting AI decide "whether this batch of goods should be shipped," it's at a loss. Why? Because this requires understanding complex contexts such as customer relationships, market environments, and brand positioning, which are precisely the advantages of humans.

#3

Transparent Society: Privacy Is Dead, Co-Veillance Lives Forever

The most controversial view in K.K.'s book is about privacy. He straightforwardly says that to enjoy the convenience of the AI era, you must give up privacy. You can't have your cake and eat it too. Want an AI assistant to arrange your schedule? It needs to know all your itineraries. Want personalized medical care? It needs your genetic data. Want accurate recommendations? It needs to analyze your every move.

But K.K. proposes an interesting concept - "co-veillance." Simply put, it means: You watch me, and I watch you; if you collect my data, I also need to know who you are, what you've collected, and what you're using it for. It sounds great, but the reality is that the party that controls the data is always reluctant to be transparent. Just like a casino owner will never tell you the algorithm of the slot machine.

However, K.K. has an interesting suggestion for China: "If China can build a truly transparent system with co-veillance, it will become the ultimate information-based country." In this system, corruption will be identified in real-time by AI, policy-making will be based on data rather than gut feelings, and everyone can enjoy customized public services.

The ideal is beautiful, but the reality is harsh. But at least this direction is correct.

#4

Sino-US Game: The Possibility from Zero-Sum to Positive-Sum

Another highlight of the whole book is K.K.'s thinking about Sino-US relations.

He starts by presenting a very subtle metaphor that we seem to have heard many people quote: The Sino-US relationship is like a bad marriage. The two sides have fallen out, but they're reluctant to divorce for the sake of the "child" (the global economy). They don't trust each other, but they can't do without each other. China needs the US market, and the US needs Chinese manufacturing. Both sides are making compromises, paying a price, and wanting to "divorce" but can't separate. What's even more interesting is that there's a "third party" in this marriage - India.

Then, K.K. proposes the concept of the "New Chimerica," which is more optimistic than Niall Ferguson's "Chimerica." This concept is far more complex than "US innovation, Chinese manufacturing." His core insight is: Innovation requires chaos, and scaling requires order. Precisely because of this contradiction, China and the US can form a peculiar complementary relationship, especially in the high-tech field. His basic logic is as follows:

• The US's strength lies in breakthrough innovation. To achieve this, you have to tolerate failure, challenge authority, and break the rules. This will bring social costs - income inequality, crime rates, and instability. The US says, "It doesn't matter; this is the price of innovation."

• China's strength lies in large-scale application. To achieve this, you need to execute efficiently, collaborate precisely, and optimize continuously in an orderly environment. China has a natural advantage in this regard - strong organizational capabilities, safe cities, a complete industrial chain, and a huge market.

In the future, millions of Chinese people will go to the US to do innovation every year, and at the same time, a considerable number of Americans will come to China to do application. "If you want to do crazy things, go to the US; if you want to promote your product to over a billion people, come back to China." K.K. thinks this is the best way to save the "Sino-US marriage."

But I think K.K. underestimates two factors:

First, China's innovation ability is rapidly improving. BYD's blade battery, CATL's Qilin battery, and Huawei's 5.5G - these aren't just simple "from 1 to 100" developments; they're real technological innovations. China won't always be content to be a contract manufacturer.

Second, the US's "innovation advantage" is being eroded by its own political correctness. When Silicon Valley companies focus more on DEI (Diversity, Equity, and Inclusion) than on technological breakthroughs, and when the US's STEM education is hijacked by various ideologies, this innovation advantage will also be eroded.

When it comes to the Sino-US AI competition, many people compare it to the US-Soviet nuclear arms race. K.K. says this analogy is completely wrong. Nuclear weapons can only destroy, while AI is creating; the nuclear arms race is a zero-sum game, while the AI competition is a positive-sum game. The result of the Sino-US competition is to train better AI, and the whole world will benefit. More importantly, nuclear technology is closed, and the knowledge of making nuclear bombs is strictly confidential; AI is open-source, and most papers, codes, and models are public. DeepSeek was able to train a GPT-4-level model with $6 million thanks to this open-source ecosystem. So K.K. predicts that in the future, there may be "one technology, two systems": China and the US will share the underlying AI technology but develop different application systems based on their respective values.

However, all these assumptions are based on one premise: trust. And this is precisely what's most scarce at present. The US's dependence on China is "higher than ever." It's not just dependence on manufacturing but also on the market and innovation. Want to decouple? The US has to "exclude anything made in China from the supply chain," which is almost an impossible task. But the deeper the dependence, the stronger the sense of insecurity. It's this ambivalent attitude that makes the US vacillate in its policy towards China.

#5

Organizational Revolution: The Disruption of Middle Managers

K.K.'s prediction about the future organizational form may be the most easily overlooked but the most important part for corporate managers.

His core view is that AI will replace middle managers.

Why? Because the main tasks of middle managers - information aggregation, passing on information up and down, and coordination - are precisely what AI is best at. When every employee has an AI assistant, when all data is visible in real-time, and when decisions can be based on algorithms rather than experience, what's the meaning of the existence of middle managers?

K.K. predicts that future organizations will show a polarization:

• Super-large enterprises: With a scale of one million people, but the organization is extremely flat and coordinated by AI;

• Super-small companies: One- or two-person companies will spring up everywhere, and the first "super individual" with an annual sales of $1 billion is about to appear.

This reminds me of the popular "middle platform" concept a few years ago. Many enterprises spent huge amounts of money to build middle platforms, essentially trying to replace middle managers with systems. But most of them failed. Why? Because today's IT systems don't have the flexibility of AI.

But after the emergence of ChatGPT, the situation has changed. I know a friend who does cross-border e-commerce. He uses AI to manage businesses in more than 20 countries with a team of only five people. Previously, he needed operation managers in each country, but now one AI can handle it all - it can reply to customer emails in 20 languages simultaneously, analyze market data from various countries, and even automatically adjust marketing strategies according to local festivals.

#6

Education and Healthcare: The Real Battlefield for AI Inclusiveness

K.K. believes that the biggest transformation brought about by AI isn't in industry but in the service sector, especially education and healthcare.

In education, personalization will become the norm. Every child will have a dedicated AI tutor, customizing courses according to their learning speed and interest characteristics. The AI teaching assistant of Khan Academy is already doing this - it can not only grade homework but also identify students' knowledge blind spots and provide targeted tutoring.

What's even more interesting is one of K.K.'s views: In the future, the most important skill for high school students isn't math or English but "learning how to learn." Because knowledge will become outdated, but the ability to learn won't.

In healthcare, customization is the direction. K.K. envisions a "3D pill machine" - formulating drugs specifically for you on-site based on your genes, medical history, and real-time health data. It sounds like science fiction, but WuXi AppTec in China is already doing something similar, although it hasn't reached the consumer level yet.

K.K. specifically mentions China's advantage: If China can establish a national genetic database, it will gain an insurmountable leading position in medical innovation. The genetic data of over 1.4 billion people is a resource that no other country can replicate. But the problem is: Who will manage this data? How can we prevent its abuse? This brings us back to the issue of "co-veillance."

#7

China in 2049: Cool or Not?

At the end of the book, K.K. uses a very interesting word to describe the China he expects in 2049 - "cool." What is cool? It's not about being rich or powerful; it's about being desirable, imitable, and inspiring the best in others. K.K. says that Japan became cool through anime and games, and South Korea became cool through K-POP. What about China? He offers some speculations:

• China's urban design will be imitated worldwide;

• China's games, music, and movies will become popular globally;