Original · Unique Research · 2026-04-06
Editor's note: This historical edition translates the source's full first-person report and roundtable discussion, retaining repetition and disagreements. First-person observations belong to the original writer or named speaker, not the translator. Corporate scale, revenue mix, market share, IPO ambitions, partnerships, product performance, sold-out stock, workforce changes, and learning multipliers are reported claims, not independently audited facts. The opening calls the three-million figure installations; later passages call it system downloads. The report uses OpenAI in one passage and OpenClaw in the detailed discussion, and gives the collaboration group different names; those variations are retained rather than silently reconciled. The opening attributes Token-based employee ranking to the partner's own staff, whereas the later account attributes it to a client. Both statements remain as written. The desk example combines a 1.4-meter width with a 60 × 120 cut; those figures are reproduced without repairing the dimensional inconsistency. Chinese personal and company-name romanizations remain provisional; Akon's brand is identified as xTool in his own introduction. Hiring, dismissal, asset-management, security and “hack” remarks are attributed discussion, not instructions or recommendations. Relative dates refer to the conversation, whose exact date is not established here, rather than automatically to the April 6 publication date. The 13 source images and final semantic/conflict review remain outstanding; this is not yet an approved zero-omission edition.
Extraordinary Awards
AGI Across Boundaries: Exploring the Technical Limits of General Intelligence and Its Integration into Industry
The Physical World Has Not Been Upended Yet, but People Have Already Been Transformed
"
A laser-engraver company derives 97% of its revenue from European and American markets. Now it is training AI to recognize wood, stone, and acrylic.
A home private-cloud company, with more than three million installations worldwide, is pushing forward a new computing category: the next-generation Agent Computer.
A core Google partner in Asia-Pacific has watched a logistics company use AI to eliminate two thousand customer-service jobs worldwide, then turned around and told its own employees: the more Tokens you spend, the better you are.
The founders of these three companies sat at a roundtable at Hangzhou AI WEEK and talked for an hour. They discussed AGI, hardware, global expansion, and organizational change. But the more I listened, the more I felt they were really discussing the same thing:
The physical world has not been upended yet, but people have already been transformed.
Your Household's Next Computer Will Look Like This
Pan Xinlei—Lauren, founder of Bingjing Technology—has a view that sounds like a prophecy. But he says they are already acting on it:
“The next-generation form of the laptops everyone here is taking out will not look like this. It will be an intelligent assistant in your home, online 24 hours a day.”
Bingjing's core products are home private-cloud systems—ZimaBoard and ZimaCube—which have considerable recognition among tech enthusiasts. With more than three million system downloads, its main users are in Europe and North America. It is now embedding Agent capabilities like OpenAI's in home computing devices. In his words: build a computing device centered on an Agent, then bring it into households everywhere.
I asked where this view came from. He took a detour and told an old story:
Before the iPod, people used the Walkman. At the time, the Walkman was the best medium for listening to music. You went to a physical shop to buy tapes, which was expensive. Then technology arrived: the iPod put a thousand songs in your pocket, with one-click purchases at one US dollar per song.
“What technology can change is the best solution for satisfying people's intentions and needs. But the underlying need does not change: people still want to enjoy music anytime, anywhere.”
He said the same logic applies today. People need personal computing devices to handle tasks; that need has not changed. What has changed is how the tasks are handled—from mouse and keyboard to natural language.
So what Bingjing now wants to build is not “a NAS with AI added.” It is an entirely new kind of infrastructure: a home computing center that is sufficiently trustworthy and secure to take charge of your household's digital life around the clock, while running several Agents to handle everyday tasks.
Four or Five Colleagues Live in the Feishu Group. They Are Not Human.
After outlining this vision, Lauren suddenly paused and said: “Inside Bingjing Technology in 2026, it is actually a different picture.”
We have a team collaboration group called the “Silicon–Carbon Collaboration Group.” It contains a dozen or so people, plus four or five Agents. These Agents handle all our everyday marketing, market activities, product development, and analysis of public sentiment and reviews.
“You can assign an Agent in the group called Logan to do all your communications research and user insights. It can speed up the return flow of market feedback.”
Then he added a line that made the whole room laugh:
“I can also hack it and see whether your boss has recently had you do anything they would not want the company to know about.”
This detail is not just a joke. It points to a structural change already underway: in the future, your work collaboration network may well contain “colleagues” who are not human. You can call on them to realize any idea, or have them take part in collaboration as your stand-ins.
Wu Guangyu, CEO of Baidao Data, has a more direct observation. As a core Google partner in Asia-Pacific, he serves many enterprise clients and has experienced the full process of deploying AI across industries.
He said different clients have very different AI requirements. A logistics client used AI to replace more than two thousand customer-service staff worldwide, retaining only a handful. Drug-discovery clients need AI to process vast quantities of in-depth data. Healthcare requires precise visual recognition. Manufacturers need extremely long context windows capable of absorbing tens of thousands of pages of product manuals.
“What they require from a model is a very comprehensive set of capabilities.”
He also made a somewhat disruptive prediction: the term “multimodal” will disappear.
“More than twenty years ago, there was a term, ‘multimedia computer.’ A computer with a color monitor and sound card counted as multimedia. Today, if you told me a computer could not play sound or show color, I would find that unimaginable. In the future, a large model without unified understanding may not count as a genuine AI product.”
The Physical World Has a Boundary Called the Size of a Shipping Container
Akon, a technology partner at a smart-hardware company, was consistently the most grounded voice at this roundtable. He comes from a cross-border smart-hardware team. More than 90% of its revenue comes from European and American markets, it is working toward an IPO, and its share of GMV in its vertical market is approaching 50%. That already sounds impressive. But what it is doing with AI sounds more practical and closer to users:
Put cameras into machines so AI can recognize materials.
Wood, stone, glass, acrylic—once AI identifies the material, the machine automatically matches the power and speed. Users no longer have to adjust the parameters manually, making creation easier.
Then he paused and said the longer-term goal was to eliminate the “computer” in the middle. “In the future, you will tell the machine directly, ‘I want to engrave something,’ using voice interaction. You will not need an intermediate device for the interaction.”
But when the moderator, Fiona, asked whether AI was redefining the form of hardware,
Akon did not take the bait. His answer was clear-eyed:
“There is a major difference between the physical and virtual worlds: the physical world still has many constraints.”
He gave the example of an office desk. Why are most office desks around 1.4 meters wide? The reason is simple: a common sheet size is 244 × 122 centimeters, and the most standard, material-efficient cut is 60 × 120. So desks on the market are all roughly similar in width.
“Logistics requires shipping containers, and containers have standard dimensions. Your product needs to be transported, with tiered freight charges imposing limits. Working backward, you have to design it within a reasonable volume. Peel back the layers: weight, dimensions, manufacturing processes, all kinds of base sheets… These are dimensions the physical world cannot yet transcend.”
Then he offered a judgment that left Fiona silent for two seconds:
“Software still serves hardware, not the other way around.”
This differs subtly from Lauren's line of reasoning. Lauren is saying that natural language redefines interaction. Akon is saying that supply chains and physical constraints limit change.
Who is right? I think both are right within their own frame. One sees a software-layer revolution; the other sees hardware-layer constraints. Different timelines produce different judgments. But that very disagreement shows that integrating AGI with the physical world is much harder and slower than integrating it with the digital world.
Eliminating the Lowest-Ranked Employees by Token Consumption Is Not a Joke
In the second half of the roundtable, the topic shifted to organizations.
Wu Guangyu said they had a well-known Chinese smart-appliance client whose AI requirement was that everyone use it, with employees ranked each month by Token consumption and the lowest-ranked eliminated.
That means employees who do not spend money using AI get fired.
It sounds radical. But his view is that AI's impact on organizations will have two stages:
In the first stage, some departments will disappear outright. Just as public cloud eliminated data-center operations jobs, AI may make many underlying data-analysis positions unnecessary. He mentioned that after AI is introduced, the number of people coding might fall by 80%.
In the second stage, new jobs will appear—but not the kind you imagine. Not “prompt engineers”: he sees that role as transitional, disappearing once AI has sufficiently strong general capabilities. What will endure is an “integrator” role combining an understanding of business logic with the ability to map that business onto AI.
It is something like a product manager, but harder: you need to understand the business, Agent workflows, and the limits of model training.
Akon's version was more direct: “The best collaboration is no collaboration.”
He said the company constantly simplifies internally. If one person can do something, do not use a five-person team. But that “one person” is armed with AI.
He gave the example of the design process: product manager → interaction designer → UI → frontend, originally four people. Now everyone can become a product manager, including the UI designer. The chain in the middle has been broken, but each person has become broader.
“People should move upstream. Previously, you could do only one thing; now you can span five fields. That is the direction of future career development. You are more capable than before.”
Lauren responded from another angle: models give people tremendous learning leverage. He said a new graduate who is willing to learn may learn 50 to 100 times faster than he did when he first started work.
“Models will push your learning bandwidth to the limit.”
But he said that was a good thing, not a bad one. The industrial era forced people into the role of cogs. The AI era, by contrast, is freeing them. “You want to do what interests you, and if you are willing to learn, you can be rewarded.” That utopian picture is beginning to look a little like reality.
I Do Not Want to Live in WALL-E
Finally, Fiona asked an open-ended question: what does the AI future look like in your imagination?
Akon did not offer an answer. He offered two films: WALL-E and Ready Player One.
“Humans sit in hovering chairs, constantly watching screens. They cannot even dress themselves and are fed like machines on a production line. The real world is in ruins, and they become powerful only in the game world. I feel that this is not far away.”
He said he did not want to become like that. So he insists that AI's arena should be the physical world—smart manufacturing, health and medicine, real creation—not feeling invincible because you can click on a few things on a screen.
Then he added: “Technology should serve good. Many people now use AI for fraud and to harm vulnerable groups. I do not want AI to bring us that.”
Lauren's version was a little lighter:
“I think lying back is quite nice, actually. But after a while, even that gets tiring. It is like being a student: you want a holiday, then when it arrives, you get bored. So you stand up: ‘No, I want to plant trees.’ Then you find tools to help you plant them, and that creates good value, too.”
Then he said: “The transition will be difficult. But once we get through it, humanity will always find a way.”
I hope he is right.
A Final Word
The three people at this roundtable seem to work on completely different things: home servers, laser engravers, and deploying Google's AI. Yet they are strikingly aligned: all are trying to bring AI's power into the physical world rather than leave it confined to screens.
Their differences concern speed and path. Lauren is imagining what the next-generation PC will look like. Akon is confirming that shipping containers are still the same size. Wu Guangyu is helping enterprise clients work out how many people they laid off this month and what kinds of people they hired.
I came away from this discussion with one judgment:
AGI will enter your work quickly. But to enter your life, it must still cross the physical world's supply chain. However long that chain is, that is how long AGI will take to reach your doorstep.
First, do the half inside the screen well. Then wait patiently for the other half.
More Details from the Conversation
Extraordinary Awards · Hangzhou AI WEEK Trends Roundtable Panel: “AGI Across Boundaries: Exploring the Technical Limits of General Intelligence and Its Integration into Industry”
Guests: Pan Xinlei, Founder of Bingjing Technology | Wu Guangyu, CEO of Baidao Data | Akon, Technology Partner at Chuangke Gongchang
Moderator: Fiona, Co-founder and COO of CGL
Fiona: Thank you so much, because all three guests have traveled a long way. I do not know whether many people here today are local to Hangzhou. Akon—Mr. Cai—is a partner at Chuangke Gongchang and has come all the way from Shenzhen. Mr. Wu also came from Shenzhen. Lauren drove down from Shanghai especially this morning, so everyone has traveled a long way. Let us take this opportunity to have our three guests briefly introduce themselves and their companies' businesses. Lauren, please start.
Lauren: I am very happy to share in this panel. I am the Founder of Bingjing Technology. We make a home private-cloud product. Of course, recently we have been integrating OpenClaw-like capabilities into the household Home Agent. Over the past two or three years, Bingjing has had more than three million system downloads worldwide and is the largest Chinese provider of core home private-cloud software expanding overseas. Our main users are in Europe and North America.
The future we see is one in which every household has a Home Agent. Like most hardware and technology enthusiasts, you will have a household “Jarvis” to take care of all your daily life and meet your needs in productivity scenarios. It will be an integrated hardware–software solution that is local, trustworthy, and secure. These are some of the business directions Bingjing is pursuing.
Wu Guangyu: Hello, everyone. My name is Wu Guangyu, and I am from Baidao Data. Baidao Data is a core Google partner in Asia-Pacific. Our business is dedicated to bringing Google's full-stack AI capabilities to our customers.
As everyone knows, AI is essentially part of Google's DNA. Among the major tech companies, Google has the most comprehensive, full-stack AI capabilities: from underlying computing power to AI frameworks such as TensorFlow, on which we can build our own models; then Google's open-source models and commercial offerings such as Gemini, Veo, and Nano Banana; and higher-level tools and capabilities such as Vertex's Agent Space, which help you build your own AI systems. We use Google's full stack to help all our customers meet their AI needs.
Akon: Hello, everyone. I am Akon from Shenzhen Chuangke Gongchang. Our brand is xTool. We focus primarily on laser engraving and hope to enable more ordinary people to use industrial-grade products in everyday life, lowering the barrier to creation.
We really have two key phrases: one is “combining hardware and software,” and the second is “going global.” Our company derives 97% of its revenue from European and American markets, so going global is central to who we are. I was just discussing this with the other guests: we have always felt that, as AI arrives, more of our effort should be directed toward the physical world. Many problems in the physical world remain unsolved, so there is no need for everyone to be especially anxious. By using more AI knowledge when we interact with our lives and the everyday things around us in the physical world, we can gain more satisfaction or emotional value.
Fiona: Thank you to all three guests. We have spent the morning discussing pure AI, AI-native software and data, how to handle contextual understanding, and good technical solutions. In this segment, our three guests all have two important characteristics: deep integration with hardware, and global expansion. Those traits are very much part of all three of your companies.
My first question is this: different smart-hardware products—what we might have called smart hardware in the previous era—have different needs on the hardware side. How will they combine with AGI in the future? For example, how are your three companies putting this into practice? Mr. Wu, you could also share some good examples you have seen of AGI being applied to enterprise products.
Wu Guangyu: I feel fortunate that we were able to enter AI through public cloud. Because we serve many customers, we have also seen a great variety of AI application areas.
In terms of how customers apply AGI, I think there are two aspects. One is using different AI capabilities. For example, we encountered a logistics client that used AI to replace a large customer-service workforce of more than two thousand people worldwide, retaining only a small number. So it needs very strong semantic understanding, translation across languages, and contextual communication. We also see globally known drug-discovery companies that need to analyze and understand vast quantities of in-depth data to support drug development. In healthcare, video and image analysis must be precise. Manufacturing clients, meanwhile, may require very long context windows: for example, a manufacturer might have tens of thousands of pages of product manuals that AI needs to grasp and understand fully. So different customers need different characteristics from AI.
The other aspect is how AI is presented. Some clients want to converse with it through a chatbot. Others embed AI in their apps. With still others, you barely notice the AI at all because it is embedded in underlying services such as databases. Taken together, we see that what they require from models is a very comprehensive set of capabilities.
One major topic we used to discuss often was “multimodality.” But I actually think the term “multimodal” may disappear or become obsolete in the future. I do not know whether everyone remembers this, but more than twenty years ago there was a term, “multimedia computer”: a computer with a color monitor, sound card, and speakers was called a multimedia computer. Today, if a computer could not play sound or did not have a color monitor, I would find that completely unimaginable. So in the future, a large model without unified understanding, without the ability to understand multiple modalities, may not qualify as a genuinely good AI product.
Fiona: Yes. Lauren, what is your view?
Lauren: Our category is fundamentally the PC category. We know that, worldwide today, after the little lobsters—large-model applications—took off, the Mac mini must surely have sold out. But Bingjing sees that the Mac mini is not actually the best computing device or medium. In other words, the next-generation form of the laptops everyone here is taking out will not look like this. It will be an intelligent assistant in your home, online 24 hours a day. An entirely new kind of Infra will emerge as a home computing center entering households everywhere. It should be sufficiently trustworthy and secure to take care of all your household data, while bringing as much home-assistant capability into your home as possible. A kind of agent computer will emerge: a computing device centered on an Agent. That is a core direction Bingjing is pursuing as we ship products globally.
Back at our company, though, it is a different picture. I think there are two very new categories here, with tremendous room for imagination and action. In our office, where many colleagues use Feishu, we have a collaboration group called “Silicon-Based Explorer.” It may contain a dozen or so people plus four or five Agents. I can hack into my colleague's Agents and use them to address my own productivity needs. I can also hack one to see whether its boss has recently had it do anything they would not want the company to know about. [Laughter.] These four or five Agents are already handling all our everyday promotion, Marketing, product development, and the latest YouTube reviews. You can assign an Agent in the Feishu group called Logan to do all your communications research and user insights. It accelerates the return flow of market feedback. Fundamentally, when you are a Chinese company operating globally, the information gap is fairly large. It can speed up the flow of global information back into your team.
So extending that reasoning, another category appears: an agent solution for what we call SMBs, or what we call OPCs.
Akon: There are two key dimensions in our company: internal operations and products.
Internally, as a company gradually grows, collaboration becomes especially important. We firmly believe the best collaboration is no collaboration. If you can finish something yourself, do not bring in five other people to do it with you. So we constantly simplify our business processes and management. If you can handle something on your own, do not tell me you need five teams. How do you solve it? OK, use AI, Agents, and so on. We find that when tasks are assigned, information degrades. After being passed along five times, it turns into something else, then comes back around for more changes. That is the key factor affecting efficiency. So I think there is much more room to apply and iterate on AI internally.
Second, hardware products. Because we make smart hardware and laser engravers, they contain many intelligent features. Let me give two examples. Engraving differs by material, so we have built-in cameras. With AI, the machine can tell whether something is wood, stone, glass, or acrylic. It can then automatically set power, speed, and other parameters, lowering the barrier to creation further. Looking ahead, our current interaction still requires a computer. But in the future, I might tell my little machine—say, a 3D printer—“I need to engrave something,” directly eliminating the computer in between. It might be voice interaction alone. I think these are changes and more efficient ways of collaborating that AI can bring us.
Fiona: I would like to follow up with both of you. People are very curious about hardware. Previously, the hardware's form defined how its software would deliver an outstanding user experience. But many people now talk about the reverse: starting from something AI-native, I define what form the hardware product should take in the future. The underlying logic changes fundamentally. As AI continues to iterate, are your companies thinking about following this trend to bring an entirely new product into being? Or overturning the original product definition and rebuilding it? Is that happening?
Akon: Well, there is a major difference between the physical and virtual worlds: the physical world has many constraints. For example, logistics imposes constraints. Train carriages are always that size, and freight containers are always this size. Work backward, and your product can only be made to fit those dimensions. Peel back the layers: weight, dimensions, and the template dimensions in the manufacturing process chain. Here is a little-known example: why are everyone's office desks 1.4 meters? Because furniture sheet dimensions are 244 × 122, so most desks are cut to 60 × 120. Physical constraints impose many limits you cannot exceed.
By that logic, software serves hardware, not the other way around. Unless we develop into the next industrial era—for example, when industrial robots arrive and we are no longer constrained by these things—it might then reverse. For now, I still think software necessarily serves hardware.
Fiona: Understood. So can I take that to mean that the whole industrial chain is not yet that advanced? After all, the physical world is still constrained by upstream and downstream supply chains. If one day robotics is more developed or the supply chain becomes more flexible, allowing new forms to be quickly created through DIY, then that limitation might be broken. Is that a fair understanding?
Akon: That is how I understand it based on what I currently know.
Fiona: Yes. Let me turn the question back to Lauren. I know you have been extremely busy every day since January this year. On one hand, you are redefining the form of future products and have spotted an extraordinary opportunity. On the other, I hear many investors are seeking you out. What opportunity do you see, and in what direction will the changes and adjustments you make take you?
Lauren: Following Akon's line of reasoning: in the category we work in, how do you distinguish noise from a real product capable of creating human value at scale? There is a very simple guiding thread.
Take an example: before the iPod, people used the Walkman. At the time, the Walkman was the best medium for enjoying music, but buying tapes in a physical shop was expensive. Technology changes the best solution for satisfying people's intentions and needs. So the iPod put a thousand songs in your pocket, and one click could earn one US dollar for a song selection. You can work only along the trajectory of technology, but the underlying need does not change: people want to enjoy music.
That means, returning to our “household Jarvis” or next-generation PC agent computer, people's unchanging need is for a personal computing device that handles tasks. Originally, you typed and used a mouse and keyboard. Today, OpenClaw has opened the door to an entirely new form of “mouse.” You can take out your phone and use natural language to ask the three or four Agents at home to handle everyday needs, from paying bills to having them balance your household's asset reserves when conflicts arise. The mode of interaction has changed dramatically.
Some people reason in terms of “AI-native products,” while another camp talks about “adding AI capabilities to an existing category.” I think that is a matter of wording. Fundamentally, the logic of a product company does not change. For example, Akon's team studies how to make CNC sufficiently convenient and easy to use for people's creative work; the human need for expression does not change. Once you identify the unchanging element and find the best technological solution along that path, you can certainly build a high-quality product with global impact. That is how we view it.
Fiona: Understood. I find that especially interesting. We have been discussing products and your thinking about future changes in AI applications, and along the way have touched on organizations and people. We focus on organizations and people every day—that is our core business—so I am curious: has AI changed the form of your organizations? Has it changed how you define people's roles? Have you considered layoffs, feeling that Agents can already replace labor and that fewer people would be better? Mr. Wu, please start.
Wu Guangyu: Certainly. I actually have quite a lot to share here. We have seen many interesting changes among our clients in the AI era.
Let me start with a personal experience. At a family gathering this Lunar New Year, I told everyone: regardless of age or gender, retired or not, everyone must embrace AI. If you do not use it, you will certainly fall behind in the workplace. Before the holiday, a very well-known Chinese smart-appliance client required everyone to use AI, ranked employees each month by Token consumption, and eliminated the lowest-ranked. It became a case of the more you spend, the better you are; if you do not spend, I get rid of you! They went further and required AI to enable 100% natural-language coding.
As for social organizations, I think the process of change will have two stages.
In the first stage, some departments in many companies may disappear. Just as many people resisted public cloud when it first appeared, only for all the data-center administrators—the operations department—to be eliminated after adoption. Now, with AI, the number of people coding might fall by 80%. For some enterprises, once the underlying data comes in, frontline business staff can analyze it directly through AI, and a large underlying data team may no longer be needed. So many departments really will shrink or be abolished.
On the other hand, some jobs will be added. When computers first appeared, there were dedicated computer operators. Today, when recruiting, we look at whether you understand AI. Prompt engineering exists because AI's general capabilities are not yet strong enough. If they become strong enough in the future, prompt engineering will no longer exist. So in the short term, AI administrator and Agent administrator roles will emerge.
Beyond that, we need a role similar to a product manager: someone who understands business logic and how to match it to AI, Agent workflows, and the constraints of model training. At the underlying level, some AI engineering and development companies will take on customization requirements. These will be enormous changes within enterprises.
Akon: I strongly agree with what Mr. Wu just said. Earlier, I mentioned that we are simplifying enterprise collaboration. When collaboration truly reaches the point of “no collaboration,” what should the other colleagues do? This is actually a revolutionary capacity for self-iteration.
More than ninety percent of people are resistant to change. How do you counter that? Through a personal update. Take a simple example: the software industry used to have a job called “graphic artist,” which later became “UI designer,” followed by “interaction designer.” The old workflow was product manager → interaction designer → UI designer → frontend, requiring four people. But since last year, people have found that everyone seems able to become a product manager, breaking the chain in the middle. The product manager no longer needs interaction and UI designers to help think things through. So what does the former UI designer do? The UI designer can become a product manager, too.
So I think people should move upstream. Can today's product manager take on end-to-end interaction? Previously, interaction meant the person interacting with a screen. Now it should begin the moment the product is unboxed—end-to-end interaction across hardware and software. Previously, you could do only one thing; now you can span five fields. That is the direction of future career development. You are more capable than before.
Lauren: Following Akon's reasoning, our organizational cultures certainly share some similarities: we both like experimenting and tinkering. Bingjing sees the same picture Akon describes. That line works well: “If one person can do it, do not have several people do it.”
Models give us tremendous leverage. We used to think new graduates had little room to maneuver. But we have found that if someone fresh out of university is willing to learn and enjoys doing something, their learning-leverage factor may be 50 to 100 times what mine was when I first started work. Models push your learning bandwidth to the limit. For example, if my daughter is interested in the Terracotta Warriors, she can work through the whole subject, from their origins and history to excavations in different periods.
This creates an effect within the organization: there is always a group of people who like to build something, not simply for work. When they have that intention, they look for tools and learn; once they build something, they receive a reward, completing the reward loop. For people willing to embrace new things, productivity leverage increases exponentially.
Taking the reasoning further, I think this is a good thing overall. Industrialization and Taylor's management philosophy forced people into the role of cogs, turning work into segmented SOPs. Now a utopian world in which “you want to do what interests you, and if you are willing to learn, you can be rewarded” is becoming reality. But there will be painful adjustment during the transition.
Fiona: This is a very interesting topic, but time is limited, so let me move quickly to the next one. This is a rare technological revolution, changing life, education, consumer behavior, and work. I am curious what the future looks like in each person's mind. Where can AI lead us? Will it make people more humane, or more boring? Will it free everyone from a life of drudgery to pursue what interests them, or will many simply choose to lie back and opt out? What picture do you imagine with AI?
Akon: I may not know what the future or the best outcome is, but I want to make clear what I do not want us to become. There are two films everyone has probably seen: WALL-E and Ready Player One. The real world is already in ruins; WALL-E is alone on Earth collecting rubbish, while humans have all become enormously overweight, sitting in hovering chairs constantly looking at screens. They cannot even dress themselves and are fed like machines on a production line. Real society is in decay, and they become powerful only in the game world. I do not think that feeling is particularly far away. In this era of software development, we could become like that very soon.
So I insist that we should use software and AI to empower the real, physical world—in smart manufacturing, people's health, medical care, and safety. Not treat it like a god and feel invincible because we can click a few things on a computer. Second, technology should serve good. Many people now use AI for fraud and to harm vulnerable groups. I do not want AI to bring us that. Each of us must hold to a baseline of values when using it. I do not want the future to look like those films.
Lauren: To put it simply, I actually think lying back is quite nice. First, lying back is nice. But after a while, even that gets tiring. It is like being a student: you want a holiday, then when it arrives, you get bored. After lying around for a while, you start wanting to do something. So somehow, I think people will solve this problem. You eventually stand up: “No, I want to plant trees.” Then you find tools to help you plant them, and that creates good value, too. More realistically, the transition will be difficult. But once we get through it, humanity will always find ways to solve things.
Fiona: Good. Thank you especially to all our guests. People keep joking that we may be the last generation who can write essays, the last who can make PPT presentations, and the last who need to go through nine years of compulsory education and university before seeking work or starting businesses. Perhaps none of that will be necessary in the future, but I do not think that is what matters.
Which direction will the future take? One thing I firmly believe is that AI really does offer us a rare opportunity. Every technological revolution is an important opportunity for wealth redistribution. I hope everyone here can seize this revolution and benefit from it. Of course, I also hope that with AI, we can become more humane and life can be less boring—with more poetry and distant horizons beyond making money.