Dialogue with Bill Gates: The current enthusiasm for AI far exceeds the Internet bubble and will experience two more scale upgrades

On July 4th, according to foreign media reports, Bill Gates, co-founder of Microsoft, recently appeared on the popular podcast "the Next Big Idea" to discuss the future of artificial intelligence and his ideas on Superhuman AI and technological advancements.

For the past half century, Gates has been a key figure in the global tech industry and has shown remarkable foresight in predicting future trends. In 1980, he predicted that there would be a computer on every desk in the future. In the latest episode of "the Next Big Idea" podcast, he predicted that in the near future, everyone will have a personal AI assistant.

In this episode, host Rufus Griscom, along with Gates, and authors of the book "AI First," Andy Sack and Adam Brotman, discussed the profound impact of artificial intelligence in healthcare, education, productivity, and business. They not only analyzed the risks that AI technology may bring, but also looked ahead to the potential of AI in treating diseases, inspiring creativity, and ushering in an era of abundant resources.

Below is the full transcript of the conversation:

Host: Welcome to "the Next Big Idea" podcast. Today, we will explore the path of artificial intelligence, superintelligence, and what this means for all of us with Mr. Gates.

Every moment in human history may seem crucial and uncertain, as if anything is possible. However, it is undeniable that certain moments are more critical than others. We are currently in such a moment. We have witnessed the profound impact of technological change. The internet has accelerated the world's development, and now almost everyone has a smartphone that can access social media, leading to divided communities and increasingly extreme political stances. Now, we are at the beginning of the artificial intelligence revolution. What will happen in the next decade? I can't think of anyone better suited to answer this question than Bill Gates, co-founder of Microsoft and global philanthropist. For the past fifty years, Gates has been driving the development of machines that empower humanity, starting with his mission to "put a computer on every desk in every home."

He drove the development of personal computers in the 1980s, advanced the internet in the 1990s, and in recent years has been leading the fight against malaria and other diseases. In the past few years, he has been actively involved in the forefront of Microsoft's collaboration with OpenAI and the development of GPT. You may wonder why Gates is participating in this episode of "the Next Big Idea"? Over the past few months, I have been reading a book serialized by Harvard Business Review called "AI First." This book contains interviews with Reid Hoffman, Mustafa Suleyman, Sam Altman, and Gates. They all believe that artificial intelligence is not overhyped, but rather underestimated. We believe that inviting the co-authors of this book - tech expert Zack and former Starbucks Chief Digital Brotman, along with one of their interviewees Gates, to participate in this program would be very interesting.

What are Bill's views on the AI revolution? Superintelligence is on the horizon, and we seem to have no clear way to slow it down. The technologies available today have already changed the game. This is largely a positive thing. We can use AI to address the biggest global challenges we face. In the coming decades, we may live in a prosperous world. This requires us to stay vigilant and ensure that it becomes the world we and future generations expect. Mr. Gates, welcome to this episode of "The Next Great Idea."

Gates: Hello. It's an honor to be on the show today.

01 The Journey of the AI Revolution and Digital Transformation

Host: We have witnessed the journey of digital transformation over the past 40 years. Not only have you witnessed these changes, but you have also played a key role in driving them forward. You mentioned last September that the GPT-4 demo you saw was stunning. Is it more impressive than the first demo of the graphical user interface you saw at Xerox PARC in 1980?

Gates: I would say yes. I mean, I had seen graphical interfaces before Xerox PARC. It was a revelation that helped inspire Apple and Microsoft to do a lot of work in personal computing over the next decade. But compared to the new intelligence that can read and write, the impact of graphical interfaces is obviously smaller. It's already very impressive.

Host: I understand that your exploration of AI did not start recently. Going back to the 1970s when you were still a student, you had a strong interest in artificial intelligence. You wrote a letter to your parents expressing your concerns: "Dear Mom and Dad, if I choose to start this company, I might miss the revolution of artificial intelligence." That company later became Microsoft. The development of AI has been a bit slower than you initially expected, but now it is in full swing. What sparked your curiosity about AI in those early days? Has the current AI revolution met your expectations from back then?

Gates: Of course, anyone writing software is thinking about what human cognition can achieve and comparing it. When I was in high school, there were projects like the Shake robot at places like the Stanford Research Institute, which could reason, formulate plans, figure out how to move up a slope, climb a slope, and pick up blocks. Some of the key abilities, whether it's speech recognition or image recognition, felt quite easy to solve Human cognition is very complex, and technology is constantly improving in its attempt to replicate human cognition. Later, through machine learning technology, humans were able to give computer systems human-level visual and speech recognition capabilities. But even after these advancements, I have always emphasized the importance of the ability to read and express knowledge like humans. Later, language translation emerged, but this is still a very special field. However, with the emergence of GPT-4, it far surpassed the performance of GPT-3 in a very profound way. The continuous improvement of this ability indicates that artificial intelligence can acquire and express knowledge. And the entire process is very smooth, although not very accurate, it has indeed surpassed human capabilities.

Host: Indeed shocking. We never anticipated that moving pieces on a chessboard with a machine would be harder than becoming a better chess player than Kasparov. But seeing the challenges eventually unfold is quite interesting. As you mentioned, the demonstration at the Xerox Palo Alto Research Center has set the tone for Microsoft's development for possibly the next 15 years. The development of Windows and Office is a manifestation of this agenda. Do you think the progress in the field of artificial intelligence today will set new directions for development in the coming decades and even longer?

Gates: This is definitely the most important development trend at present, and it will shape the future of humanity in a dramatically way. At the same time, we are also witnessing the development of synthetic biology and robotics driven by artificial intelligence. Therefore, we cannot ignore the development in these areas. But undoubtedly, artificial intelligence will be the leading force driving change.

02 Headphones and Smart Glasses Will Become Important AI Hardware Devices

Host: In 1980, you had a burst of inspiration. You predicted at that time: "In the future, every home, every desk will have a computer." What do you think a similar prophecy for the era of artificial intelligence would be? Will we have an artificial intelligence advisor in every ear in the future?

Gates: The specific form of hardware is not so crucial, but headphones - which can enhance audio, eliminate noise, and improve sound quality - will undoubtedly become a very important product form. Similarly, smart glasses that can project any video into your field of vision will be a new form of human-machine interaction. The personal intelligent entity that I have been exploring for years will surpass human assistants because it can track and read all the content you want to read, always provide assistance, and understand enough contextual information. Today, we still do not trust software to sort emails because of a lack of understanding of context - such as, what to do next? How are these messages related to tasks? This form of sorting exists in a clumsy chronological order. Machines still organize information in a different way than humans: emails are arranged in chronological order, information is also arranged in chronological order, and social networks are arranged based on time. But now, machines can read our information like a white-collar worker. The interactive interface may become dominated by intelligent entities, such as assistant intelligent entities, psychological counseling intelligent entities, friend intelligent entities, girlfriend intelligent entities, expert intelligent entities. These intelligent entities will all be driven by artificial intelligence Moderator: It seems that the deeper the artificial intelligence assistant understands us, the greater its help will be. I imagine that in the near future, the four of us may all face a question: whether we are willing to turn on the audio so that our artificial intelligence assistant can listen to our lives throughout. I think this will bring many benefits because we will receive valuable guidance and advice. What do you think about this? If invited to turn on the audio, would you choose to do so?

Gates: In fact, computers can now read every email I send, and digital channels are also recording all my online meetings and phone calls, so you have already revealed a lot of personal information in the digital system. Therefore, if the artificial intelligence assistant can help me summarize meetings or assist me in following up on matters, the value it can provide will be enormous. The artificial intelligence assistant will have different operating modes based on different types of information. Intelligent agents have different models and partitions when dealing with different information. But for a personal assistant intelligent agent, you will put almost all information in this partition.

Andy: Since our last interview with you on "AI Comes First," I have been thinking about a question. You compared your experiences at Xerox Palo Alto Research Center with the feelings of using ChatGPT at that time. I think you have a deep understanding of computer technology and rich experience in building businesses and the impact of computers on humans, which puts you in a very unique position. I am curious, if you think the impact of ChatGPT even exceeds your time at Xerox Palo Alto Research Center, what insights does this have for your consideration of the future life of grandchildren, and what advice do you have for the next generation of leaders in dealing with the unique challenges brought by artificial intelligence? I am very interested in this perspective.

Gates: Of course, compared to other technologies, the development speed of artificial intelligence technology is faster, and its upper limit of potential is also more difficult to predict. This technology will reach superhuman levels of capability. Although we have not reached this level today, many new jobs are introducing metacognitive levels due to reliability constraints. If used properly, it will solve the common problem of talent scarcity in the white-collar field today, and over time, this impact will also extend to the blue-collar field. Therefore, this is indeed a huge milestone. Although some past experiences are helpful in this regard, it is novel enough that no one has truly faced these issues. We may also face some new policy issues, but most of these issues are very positive, especially in improving white-collar labor productivity.

Andy: What excites you the most about this new invention?

Gates: Among the various shortages currently existing, no organization feels the shortage of white-collar talent more than the Gates Foundation. We focus on health issues in sub-Saharan Africa and other developing countries, where there is a lack of teachers who can teach in the local language. Therefore, using mobile phone infrastructure, even in very poor countries, can achieve quite extensive coverage. This infrastructure can provide medical advice, act as a personal tutor, because it contacts you in your language and semantics These countries are unable to organize large-scale training. With artificial intelligence technology, users only need to pick up their phones and listen to what it says. This is very exciting for people in developing countries who have to deal with tragic resource shortages.

Host: You have been working at the Gates Foundation for 20 years, focusing on solving global health, education, climate change, and other issues. Do you think artificial intelligence will be an accelerator, allowing us to accomplish the work of the past 20 years in 5 or 10 years, or how significant do you think this acceleration is in these areas?

Gates: For some tricky diseases for which we currently do not have good solutions, artificial intelligence will provide tremendous help. In the past 20 years, we have achieved amazing results, halving the global child mortality rate from 10 million to 5 million per year. This was mainly achieved by making certain vaccines cheaper and ensuring they reach all children worldwide. That was a relatively easy goal to achieve. Now we are facing more difficult problems. But with the help of artificial intelligence, we can explore further upstream, such as why children are malnourished, or why it is so difficult to develop an AIDS vaccine. We can be more optimistic about these major breakthroughs. Artificial intelligence will help us in every aspect of these things, whether it's recommendations, delivery, diagnosis, or scientific discoveries, all advancing at an amazing speed, and the Gates Foundation is very actively involved in funding this work.

03 AI Scale Law Still Applies, with Two More Scale Increases

Host: Yes. We recently invited your friend Sal Khan to the show and had the opportunity to delve into the AI teaching assistant Khanmigo, I was very impressed by its capabilities. I know you recently visited schools in New Jersey to see the implementation of the new Khan Academy project. The idea is very exciting, aiming to massively improve education, with an impact on billions of people that is immeasurable.

Gates: Yes, Sal Khan's books do not explain what kind of world we are educating children for. If artificial intelligence can be fully applied in education, it would be a miracle, because there are other factors changing at the same time, making it more complex. But in the field he proposed, if education can fully utilize artificial intelligence, it would be extremely positive.

Host: Indeed. This leads to a personal topic. You recently had another granddaughter born. I also know that Adam has a 7-year-old child. When we think about this issue, we wonder what the future will be like. I am glad that our children will have a private tutor of Aristotle's level to further accelerate their education process. But the question is, what do they need to know to be effective in the world?

Adam: That's interesting, I wanted to talk about this issue from a slightly different perspective. My daughter can truly see me using the Whisper model on ChatGPT (a universal speech recognition model developed by the OpenAI team). She witnessed me living in a world of artificial intelligence. Watching her use voice interfaces so comfortably, especially at her age, is fascinating. For her, using a voice interface is actually easier than learning to spell, after all, she has just learned to read So I think this is an interesting observation. I believe it's not just about natural language chat, but even a comparison between voice chat and click operations.

But Bill, I was going to ask you a question about direction. This direction may be a bit different, a bit controversial, maybe the word is used too heavily, about whether all these cutting-edge or foundational models have already converged around GPT-4 to form a benchmark? Some people think we may be reaching a plateau. But most of the smart researchers I follow still believe that the Scaling Law will still apply at least in the next few years. What is your position in this discussion? Do you find yourself hoping it reaches a plateau, or are you emotionally neutral due to some concerns about the technology? I would love to hear your thoughts on this.

Gates: The real frontier is not about scale. We may have two more scale increases, by acquiring video data and being very good at synthesizing data, we may be able to double the scale again. This is not the most interesting dimension. The most interesting dimension is what I call metacognition, understanding how to think about problems in a broad sense, stepping back and saying, "Okay, how important is this answer? How can I check my answer? What external tools will help me?" Today's overall cognitive strategy is very simple, it just generates each token in sequence through continuous computation, which is amazing in itself. It does not think like a human stepping back and saying, "Okay, I'm going to write this paper, here's what I want to cover. I'll put in some facts. This is the conclusion I want to make."

So when artificial intelligence encounters various mathematical problems, like Sudoku puzzles, you will see the limitations of artificial intelligence. The Scaling Law will certainly continue to be effective. But at the same time, the transition from the simple algorithms we have today to various actions more like human metacognition will change, which is a bigger frontier. It's hard to predict how quickly this will happen. I see progress in this area next year, but it's unlikely to be completely resolved in the near future. Therefore, artificial intelligence will become more predictable. Currently, in some research combining different fields, we are close to achieving high accuracy in mathematics and health fields. But open-ended questions will require a universal breakthrough in metacognition.

Host: Do you think metacognition will include establishing a feedback mechanism that allows artificial intelligence to deliberate like us humans? I heard that people like Max Tegmark, a professor of physics and artificial intelligence at MIT, have proposed a view that the ability to have a dialogue with oneself may be part of how our consciousness forms.

Gates: Yes, consciousness may be related to metacognition. Since it is not a measurable phenomenon, it is always elusive, and obviously these digital things are unlikely to have similar equivalents, but it is indeed a huge frontier area that is very similar to humans in identifying the challenges to be addressed, having confidence, and examining the way they work Andy: Writing "AI Comes First" and interviewing you, Hoffman, Ultraman, Mustafa, and others was a profound learning experience for Adam and me. I often have epiphanies from conversations with you all. I pay attention every day to the speed of technological advancement driven by many different companies, especially large ones. A large amount of funds and talent are invested in these technologies. Therefore, the speed of technological development and its potential impact amazes me, and I only have a limited understanding.

Gates: If we knew how to slow down, many people might say, "Okay, let's consider doing that." As Mustafa wrote in his book, considering things at the individual, company, and even government levels, there isn't really a feasible mechanism to achieve incentive structures. If we could understand the incentive structures at the government level, that might be enough. Those who think open source is not a problem might say, if open source becomes too successful, we might stop being open source. But do they really understand what that means? They might consider that next we might quickly discuss not allowing malicious individuals to benefit from more powerful artificial intelligence than the just. Whether in network defense, war defense, or bioterrorism defense. Although we cannot completely "put the genie back in the bottle," it does mean that some malicious individuals will gain unprecedented capabilities in a new way.

Host: So, in your opinion, in the current global environment, making the most complex artificial intelligence models open source may not be a good idea?

Gates: Yes, people generally agree on this principle, but when you try to be specific about where this principle should be applied, the situation becomes less clear.

Host: Adam and I discussed yesterday that even if we could halt the development of artificial intelligence now and maintain the status quo, it may still take ten years for Forum 3 (publisher of "AI Comes First") and others to help companies and individuals understand how to apply existing technologies.

Gates: I'm not sure about that statement because it's obvious that what needs to learn is me, not the software to adapt to me. For example, I need to learn English, which is software adapting to us, not us adapting to software. So, it won't be like before where you need to learn all new menus, files, edits, windows, help functions, etc., but you can directly tell it, "I want to do data analysis to find out which products are causing sales to slow down." It can accurately understand your intent.

So, thinking that adopting new technology will be a hindrance is not the norm. Of course, companies accustomed to old methods need to make adjustments. But if you look at areas like remote support, remote sales, data analysis, etc., just let people observe how advanced users use these tools for a week, without any manuals, just learning by example, I think in the absence of server capacity constraints, especially in affluent countries, the adoption rate will not be overnight, but it will definitely not take ten years. For example, a free product that provides arbitrary audio and text artificial translation was once seen as the holy grail of overcoming language barriers. And now, a small artificial intelligence company offers it as an additional free feature. This is really crazy Perhaps you may ask, how do people adapt to free translation services? I believe that because people generally desire to understand each other's meanings, they will not find it very difficult to accept this new service, and this process does not require a long time. For example, by separating the two audio channels with headphones, this can be easily achieved. As for the goal of service quality and covering all African languages, we plan to achieve this next year. We are ensuring that even rare and unwritten languages receive support, which requires collaboration with other institutions and the collection of relevant language data. The Indian government is also doing similar work for Indian languages. Therefore, I believe that it will not take long for people to adapt to using this service, nor will it take ten years to prompt backward regions to change their behavior.

04 Opportunities and Challenges in the AI Wave

Host: In your personal podcast "Unconfuse Me," OpenAI CEO Ultraman mentioned that they have seen up to a 300% increase in productivity among developers. In other fields, we have also seen productivity growth of 25% to 50%, which is exactly as Gibson said, "The future is already here, it's just not evenly distributed." Indeed, ensuring that all companies can fully benefit from this productivity improvement will definitely be a process that takes time. I am very interested in your comments on productivity in the first chapter of your book "AI First." You said, "Productivity is not just a measure of output per hour, it is about improving the quality and creativity of our achievements." What do you mean by this?

Gates: Whenever productivity increases, you can use the increased X% to increase the quantity of output, improve the quality of output, or reduce the input of human labor time. You are always making choices in these three areas. In some industries, people's response to increased productivity is not simply to increase usage. For example, in the tire industry, the shift from bias tires to radial tires, although the annual cost of tires has decreased by four times, people did not decide to drive four times more miles because of this. So, for some products, such as computing services or the quality of news reporting, demand elasticity is very high. If you can do better, you may choose to keep human labor time constant and focus most of the improvement on quality.

There are many other cases where this is not the case. For example, people's demand for driving mileage has not changed due to the lower cost of tires. Society is full of such things that span different ranges. So, whenever productivity increases rapidly, Microsoft internally discussed how to make the market size close to zero through efficient databases, even in highly elastic demand, there will eventually be a limit to the incremental demand.

Adam: If you were to make a prediction now, you mentioned healthcare and education, how would you answer the question about what you think the first major breakthrough application will be? For example, Andy and I often listen to a podcast where they discussed last weekend, "We have not seen any major breakthrough applications yet." This is quite interesting because I am not sure if this is correct, but let's understand it according to their statement, that we are still in the experimental stage or some other stage, that's what they mean. I am very curious about your thoughts Where do you think the first major application will appear? Just like the first important application Uber in mobile cloud positioning services.

Andy: Before that, it might be Google Maps. Maybe map technology.

Adam: Exactly. When you consider this issue, you will directly think of fields like education, healthcare, etc. Where do you think the first major breakthrough consumer application or industrial application will first appear? Why?

Gates: I think those pessimists are very creative, able to say that nothing major has happened yet.

Adam: I agree.

Gates: They don't think that conference summaries, translation work, or making product parameters more productive are major breakthroughs.

Adam: Exactly.

Gates: It's really shocking. This is the ability of white-collar work, but it should be noted that in many open scenarios, it is not as reliable as humans. You can hire humans, they may make mistakes, so you need some monitoring measures, but these AI applications, if used in new areas, are slightly less predictable because in some areas we don't push them to the limit, like in supporting phone calls or remote sales calls, you don't really push them to the edge. So I don't quite understand what they are talking about.

Andy: I think when people say this, they really know how to use their skeptical spirit, because I believe your answer is absolutely correct. The key lies in the second-order effect. Take the invention of the car, for example, it can take you from one place to another, you can even foresee the development of roads, etc. But you can't foresee the expansion of Los Angeles or the emergence of suburbs and drive-in theaters.

Using more modern examples, when the World Wide Web appeared, there were initially many websites like brochures, travel websites like Expedia, these are all ordinary first-order effects. But people see Uber as the second-order effect of technology, like they think, "You can't foresee this." Now, maybe you can foresee, maybe you can't. But that's what Adam's question is about. Look at artificial intelligence, in many ways, the rules of the game for search have changed, it has become an ubiquitous consumer activity, and ChatGPT is undoubtedly the fastest-growing and most adopted technology in history. So I am not underestimating or agreeing with those skeptics, but the real key is the second-order effect.

Gates: GPT-3 didn't seem that remarkable at the time. It was interesting enough to make some people in OpenAI feel that economies of scale might reach a tipping point. I didn't foresee that, few did. We crossed that threshold about two years ago, from a popular perspective, about a year and a half ago. So now, those who are really using it are those who are open-minded, willing to try new things.

However, it should be noted that we are only showing a part of the technology. For example, applications like image editing, it is not about teaching you how to master complex functions like in Photoshop, where you need to click 59 menus and open multiple dialogs. Instead, you just need to enter simple commands, like "remove that green sweater." But some people may doubt whether this can be done, they may find it sounds difficult This kind of gap is usually more pronounced when showcasing technology, for example, requesting to zoom in on a photo, even though the original image does not provide more details.

Or in patient tracking, artificial intelligence calls you, discusses with you, whether you have taken the prescription medicine? How do you feel? What are you doing? If people really try to understand various examples, I think they will quickly realize: "Wow, these are a lot of very specific functions." For example, when I call to inquire about tax issues or explain medical bills, the white-collar workers handling these issues often demonstrate a free and flexible attitude. Although I fully admit that in some ways, it still feels a bit like a slightly crazy white-collar worker, but we will get rid of this situation in the coming years.

Host: I know a crazy white-collar worker, he is the CEO of a rapidly growing company. He asked the top salespeople in the company what occupies most of their time during the day after sales calls? They answered that it is drafting follow-up emails. So, he created a GPT instance, integrated all their best practices and communication skills, automatically transcribed every call, and generated follow-up emails. He is now downsizing, cutting half of the sales team so that the remaining excellent salespeople can double their work efficiency.

We see a successful story: on one hand, it is an extremely efficient and impressive application of technology; but on the other hand, for the sales team being laid off, the situation is not so optimistic unless they can use new artificial intelligence technology to create a competitive company or do other things. This also raises a broader question: to what extent do we think artificial intelligence technology empowers the little guy relative to the big guy? We see that only a few large companies seem to dominate in technological development. But on the other hand, now everyone can use GPT-4 Omni for free, which is also a balancing factor.

Gates: We need to distinguish between two parts of economic activities. One is the economic activity of building artificial intelligence products, including basic artificial intelligence products and vertical artificial intelligence products. It can be said that the barrier to entry into this field is currently very low, we are in a frenzy, someone has raised $6 billion in cash for a company, and many other companies have raised hundreds of millions. It can be said that never before has so much capital been poured into a new field, and it can even be said to be a new frenzy field, compared to the frenzy in the early days of the internet or the automotive industry. Whether from the percentage of market value or from the valuation of these market values.

Before the beginning of this century, no company had a market value of one trillion dollars. Now we have a chip company, it does not produce chips, but is a chip design company, which has increased its market value by one trillion dollars in just 6 months. So, competition in the field of artificial intelligence is very fierce, but at the same time, many people are entering. Indeed, Google and Microsoft have the most capital, but this has not really prevented others from developing in basic capabilities or vertical areas Although the field of artificial intelligence tools is quite large, its share in the global economy is relatively small. For example, nowadays, a small chain of hospitals can compete with large hospital chains using these tools and provide higher quality services at the same or lower prices, thus creating higher value for customers.

Host: Our current polarized environment is of no help to the effectiveness of our government. As you mentioned in the podcast, in the worst case scenario, we may see polarization undermining our democratic system. Do you think artificial intelligence can help us achieve harmony? If so, how would it do that?

Gates: Artificial intelligence is a powerful tool, and we should at least consider applying it to all thorny issues, exploring where it can bring positive impacts, or where it may worsen problems. If someone wants to know, for example, where this article or video comes from? Is it from a reliable source, is the information accurate? Or in my news feed, what I see and what a supporter of another candidate sees? Try to explain to me what prompts them to lean towards that view. We hope that well-intentioned individuals seeking to bridge misunderstandings will have artificial intelligence tools to help them uncover misinformation, highlight biases, or help them understand how we can bridge our different perspectives on the world.

So, although it may sound somewhat unrealistic, like when people talk about "Oh, let's use geoengineering to address climate change," your reaction is always to think that technology might be the solution. I may be a bit too optimistic in this regard, but here, artificial intelligence is both part of the solution and, if we are not careful, could also exacerbate the problem. The replacement of blue-collar jobs came later than white-collar jobs, which is a good thing because it is not limited to any one industry, in fact, it is the more educated group that first experiences these changes.

05 The Human Meaning and Future in the AI Revolution

Host: One last question, you have said before that a future question that puzzles you is how to think about the meaning of human existence in a world where machines can solve problems better than us. Is this still a question that continues to bother you? How do you view this question now?

Gates: I don't believe that a person who has lived for 68 years in a resource-scarce environment, considering my age and the vastly different background I am in, would have the ability to imagine a new era beyond resource scarcity. So, I think this is an important question that people should seriously consider. However, I do not have a solution to this question, nor do I expect to have one.

Host: To some extent, you have already experienced life in a post-scarcity world, after all, over the years, you have not felt scarcity in your personal life.

Gates: Although I have not experienced economic hardship, as someone who has experienced the joy of success, facing global issues such as malaria, polio, or measles, I feel a sense of accomplishment: currently, the number of researchers dedicated to these issues and the research funding are very limited. Therefore, I feel that by using my resources to collaborate with the government, plan and implement solutions to ensure that no child dies from malaria or measles, this has brought me immeasurable value You are right, under the premise of financial freedom, activities that I do for fun, such as playing pickleball, are things that people can all participate in. Even if in the future machines may surpass humans in pickleball, it will not diminish our enjoyment. We will still cherish and enjoy the unique experience of participating in these activities as humans. However, the sense of accomplishment and drive to help solve scarce resources will eventually no longer exist.

Host: I see. One more question. There are rumors that you are writing a memoir. Can you reveal some related information?

Gates: Yes. We announced that the first volume of the memoir will be published in February next year, covering my life from birth to the first two to three years of founding Microsoft, around when I was about 25 years old. The book is titled "Source Code". I am currently editing this book because the deadline is approaching. We have received a lot of positive feedback.

Host: Did GPT assist you in this regard?

Gates: Actually, no. Not because I am against using this technology or for any other reason. I think, in the end, we may consider doing so. But for now, we are taking a more traditional approach in the writing and editing process.

Adam: Are you planning to write two books or three books?

Gates: Three books. So we may have to wait three years before the second book is released. One phase of the book will focus on Microsoft, and another phase will focus on my experience of donating funds.

Host: If you and Andy continue to play pickleball, maybe you will live long enough to write the fourth book.

Andy: We hope so too.

Gates: Making artificial intelligence better. We can make that the theme of the fourth book.

Host: Thank you very much for taking the time to participate in our discussion today. This conversation was very engaging