deepseek横空出世，人类的未来是什么模样？｜王局拍案20250205

Summary

In a recent episode of Wang Zhian's show, he discussed the groundbreaking AI project, Deepseek, developed by a Chinese company. This innovation has significantly impacted global tech giants and sparked debates on its implications. The project stands out because it's an open-source initiative, unlike its Western counterparts, and operates independently of chip advances. Zhian explores topics ranging from AI's role in shaping future economies and societal norms to geopolitical tensions, suggesting that AI might redefine civilization as we know it.

Highlights

• Deepseek significantly impacts US tech companies and stock markets, showcasing China's growing influence in AI. 🇨🇳
• Amid debates, some praise Deepseek's innovation, while others question its origins and implications. 🔍
• Deepseek's ability to create poetry parallels and potentially surpass human creativity, igniting discussions on AI's role in culture. ✍️
• The project challenges existing US tech policies, demonstrating adaptability beyond advanced chip requirements. 🔧
• Wang Zhian warns of potential societal disruptions due to AI advancements, emphasizing the need for human adaptability. ⚠️

Key Takeaways

• Deepseek, a novel AI project from China, has caused ripples throughout the global tech industry and Wall Street. 📈
• The project, unlike others like OpenAI, operates with an open-source model, sparking curiosity and concern. 🤔
• Deepseek's algorithm's independence from advanced chips challenges US tech sanctions. 🚀
• Zhian discusses potential shifts in societal structures due to AI, hinting at a dystopian future if humans don't adapt. 🌌
• The discourse touches on existential threats posed by AI, questioning the nature of human existence and purpose. 🤖

Overview

Wang Zhian dives into the incredible rise of Deepseek, a new AI development by a Chinese company that's shaken the US tech industry and caused a historic stock market drop. With its open-source nature, Deepseek courts both admiration and skepticism as opinions divide on its legitimacy and potential impacts.

As Deepseek stands out with its poetry-writing ability and operational independence from cutting-edge chips, it raises eyebrows and questions, especially among Western tech entities like Nvidia and OpenAI. The AI's provocative responses evoke both laughter and concern, exemplifying China's growing prowess in artificial intelligence.

Zhian paints a vivid picture of a future potentially dominated by AI, exploring the existential concerns it raises, such as job displacement, societal roles, and ethical dilemmas. He emphasizes the importance of continuous learning and adaptation in this rapidly evolving era, warning of the profound impact AI could have on humanity's future.

Chapters

• 00:00 - 00:30: Introduction and Impact of Deepseek The sudden emergence of the AI project 'deepseek' by a Chinese internet company has had a significant impact on the US tech industry and Wall Street, likened to a nuclear bomb-level event. Nvidia experienced a record single-day market value drop, losing 5,900 billion US dollars, signaling a major shift in the industry. Other tech giants like OpenAI, Nvidia, and Google were also extremely shocked by this development.
• 00:30 - 01:00: Reactions to Deepseek The chapter titled 'Reactions to Deepseek' discusses China's potential development of an advanced AI project, similar to OpenAI's, but open-sourced. It highlights US President Trump's concerns regarding Nvidia potentially supplying China with the necessary technology, reflected by his direct inquiry with Nvidia's leadership. The chapter also outlines the divided public opinion on the internet regarding the deepseek project, revealing two major opposing viewpoints.
• 01:00 - 02:00: Controversies and Censorship The chapter titled 'Controversies and Censorship' discusses the differing opinions on the technology called deepseek. Some view deepseek as impressive for achieving technological advancements despite a chip embargo. However, there is skepticism within the simplified Chinese-speaking communities, with accusations of it being a knock-off that has plagiarized technology, notably the distillation technique, which involves reprocessing OpenAI's data. Furthermore, when deepseek first became available, it was often questioned on sensitive political topics, such as inquiries about notable figures in Chinese history.
• 02:00 - 04:00: Deepseek's Poetic Capabilities In this chapter, the text discusses Deepseek's responses to politically sensitive questions in China. It highlights an instance where Deepseek truthfully attributed the deaths of millions to Mao Zedong, an act which went against China's Propaganda Department directives, leading to the question being deleted. Additionally, Deepseek's inability to engage with questions about Xi Jinping's lifetime rule and China's political system is noted, illustrating the constraints placed on it compared to more open AI models. The chapter reflects on the limitations imposed on AI in politically restrictive environments and compares Deepseek's functionality to OpenAI, pointing out its limitations.
• 04:00 - 05:00: Understanding Deepseek's Development The chapter discusses the censorship surrounding the name Wang Zhian on China's internet, highlighting that it cannot be found on platforms like Baidu Baike. It mentions online inquiries about Deepseek's view on Wang Zhian and notes that initially, Deepseek provided candid responses, describing Wang Zhian's past as a reporter for China Central Television, where he produced many influential programs.
• 05:00 - 08:00: Historical Development of AI This chapter discusses the historical development of artificial intelligence, focusing on instances where AI has been involved in creative tasks such as writing poetry. The transcript mentions a specific example of AI-generated poetry that was so well-crafted, it surprised the individuals who interacted with it. The poem in question contained vivid imagery and emotional depth, indicating AI's potential in contributing to the arts and humanities.
• 08:00 - 10:00: Comparison between Deepseek and OpenAI This chapter discusses a poem generated by AI without specific conditions, praised for its poetic imagery. It references Fan Pang, an Eastern Han Dynasty official, who was persecuted due to his opposition to the eunuchs, drawing parallels to a loyal minister opposed by corrupt factions.
• 10:00 - 12:00: Implications for US-China AI Relations The chapter explores the symbolism behind a quote, 'Chu guest without garments makes ghosts and gods weep,' which ties to the author's own experiences and perspectives as an expatriate and scholar from Wuhan University. This 'Chu guest' metaphor emphasizes the feeling of being unprotected or exposed when working abroad. The author speaks to the value of their work despite it diverging from China's current social development trends. They liken themselves to a 'reverse-scaled person,' someone who dares to challenge and venture into the unknown despite potential futility, symbolized by 'braving the boundless sea.' These reflections emblemize the broader implications for US-China AI relations, suggesting a push against traditional boundaries and a commitment to pursuing one's beliefs amidst geopolitical complexities.
• 12:00 - 17:00: Challenges to Human Society The chapter discusses the surprising capability of AI to compose poetry, contrasting it with the difficulty faced by humans. The author reflects on the unique challenges humans encounter when creating poetry, emphasizing that unlike mathematical problems which have definitive solutions, poetry does not have a single correct answer. This underscores the complexity and creative nature of poetic composition compared to mathematical problem-solving.
• 17:00 - 20:00: The Risk of Superhumans The chapter discusses the advanced capabilities of an AI developed by the company Deepseek, which has achieved a level of creativity in poetry writing that surpasses most humans. The open-ended nature of poetry allows for various interpretations and makes it a challenging field. The AI's proficiency in poetry is highlighted as evidence of its surpassing human intelligence in certain aspects.
• 20:00 - 31:00: Future Impact and Societal Changes In this chapter, the speaker addresses their understanding of the technical subject of 'deepseek,' despite not being technically inclined themselves. They acknowledge a lack of deep understanding but share insights and information they've gathered recently. They attempt to provide a basic introduction to 'deepseek,' tracing its origins back to the year 2016. However, the detailed analysis or technicalities of 'deepseek' are not deeply delved into.

deepseek横空出世，人类的未来是什么模样？｜王局拍案20250205 Transcription

• 00:00 - 00:30 In my opinion, over the past one or two weeks, the biggest news without a doubt has been the sudden emergence of deepseek This AI project developed by this Chinese internet company has caused a nuclear bomb-level impact on both the US tech industry and Wall Street Last Monday, Nvidia's stock in the US dropped 17% in one day, with a market value evaporation of 5,900 billion US dollars setting a record for the largest single-day drop in the history of the US stock market At the same time, the US tech industry, including OpenAI, Nvidia, and Google, were all extremely shocked
• 00:30 - 01:00 that China could actually develop such a project, not just an OpenAI project, but a truly open AI project, because it's open source US President Trump overnight called the head of Nvidia's company over asking, "Did you secretly hand over these chips to China?" Otherwise, how could China develop such an AI project? On the internet, opinions about deepseek have also split into two major camps
• 01:00 - 01:30 Some people believe that deepseek is very impressive, managing to achieve such results despite the chip embargo but many in the simplified Chinese circles, including simplified Chinese media overseas, think that deepseek is actually not good It could even be a knock-off, possibly having stolen the technology and plagiarized the distillation method The distillation technique is essentially taking OpenAI's data and reprocessing it oneself When deepseek first went online, many people asked it some political questions for example, "Who is the person who has killed the most people in China?"
• 01:30 - 02:00 Deepseek, in its naive simplicity, replied that it was Mao Zedong who was responsible for the deaths of tens of millions So, you see, deepseek violated the directives of China's Propaganda Department by speaking the truth Soon, this question was deleted Then some people asked it, "What do you think about Xi Jinping's lifetime rule? What do you think about China's system of National People's Congress representatives?" Deepseek then replied that this question is too complicated, it couldn't think it through, and asked not to be asked such questions again So everyone laughed, "Did you see that? How can this castrated version of deepseek in China be compared to OpenAI?"
• 02:00 - 02:30 Because it has censorship Soon after, I saw many busybodies online asking deepseek questions about me asking, "What do you think of Wang Zhian?" Because everyone knows that the name Wang Zhian has been banned on China's intranet It's gone; you can't even find it on Baidu Baike. At first, deepseek was relatively candid It said that Wang Zhian used to be a reporter for China Central Television and had produced many influential programs
• 02:30 - 03:00 Later, he went into exile in Japan, made many YouTube programs overseas, and was quite controversial. What it said was essentially truthful Soon, someone asked it, "Write a poem for Wang Zhian." When I saw the poem, I was somewhat shocked. Why? Because the poem was really well-written Among them, there was one poem that I thought was particularly well-written; its last four lines were written as follows: "Fan Pang has tears that cascade down the annals of history, Chu guest without garments makes ghosts and gods weep" "If you wish to ask about the resolve to brave the boundless sea, at the crest of the tide stands the one with reverse scales"
• 03:00 - 03:30 The person who had AI write this poem did not impose any additional conditions; that is, you just ask it to write a poem for Wang Zhian and it produced these four lines I must say, the poetic imagery of these final four lines is really great. Let me briefly explain it to you all Fan Pang was an official of the Eastern Han Dynasty who, due to his crackdown on the eunuchs, was later persecuted to death So, he is equivalent to a loyal minister of ancient times, who suffered the blow from a corrupt eunuch faction
• 03:30 - 04:00 "What does 'Chu guest without garments makes ghosts and gods weep' mean?" Because I studied at Wuhan University, I am called a Chu guest As a Chu guest without garments, I have now gone overseas and no longer have any protective cover What I'm doing is still very valuable in itself "If you wish to ask about the resolve to brave the boundless sea, at the crest of the tide, the one with reverse scales roams" That is to say, what I am doing is not in line with the current direction of China's social development Braving the vast sea despite knowing it is futile, that is what a reverse-scaled person does
• 04:00 - 04:30 To say that the imagery of such a poem was composed by AI in just a few seconds really shocked me at the time You have to know that writing poetry is actually very difficult for a creator It's different from solving a math problem; for math, there's only one solution path, and if you have enough computing power, it's easy to solve However, when you ask someone to write a poem, a poem actually does not have a correct answer or a standard answer
• 04:30 - 05:00 It is completely open-ended So the poetic imagery is basically for us to appreciate on our own But precisely because it has no correct answer and no standard answer, it is also the most difficult So, you see, the fact that deepseek can write poems at this level now, I think, has already surpassed 99% of people in China Even I, when writing poems about myself, cannot reach that level So in my opinion, the AI developed by Deepseek—the company Deepseek—has indeed, in many aspects, completely surpassed human intelligence
• 05:00 - 05:30 Today I will talk about my understanding of deepseek Because I am not a technical person myself Regarding deepseek, this highly technical matter, I don’t really understand it very well either These past few days I have looked at a lot of information, so I’ll briefly give everyone a simple introduction Now, this story probably has to be traced back to 2016
• 05:30 - 06:00 Everyone knows that in 2016 a major event occurred: a Korean Go player named Lee Sedol played a historic match against a program called AlphaGo developed by a company under Google In the end, AlphaGo defeated Lee Sedol with a score of 4:1 At that time, this event caused an enormous shock in the field of artificial intelligence—why? Because in earlier times, humans conquered chess with a program called Deep Blue When Deep Blue conquered chess, it relied on a brute-force exhaustive search algorithm
• 06:00 - 06:30 What is brute-force calculation? It means that in chess, every move has many possible variations Then Deep Blue, using the computer’s processing power, exhausted all possible variations and eventually defeated humans But after conquering chess, many people began to tackle Go and found that Go was nearly impossible Why? Because the variations in Go are far too many Chess has roughly 10^47 possibilities, but Go has 10^170 possibilities
• 06:30 - 07:00 10^170—what kind of concept is that? It’s many orders of magnitude larger than the total number of atoms in the universe So you cannot achieve a solution by brute-force calculation Later, Google invented an algorithm called deep learning Deep learning does not require an exhaustive search; it starts to mimic the way humans think Then it was applied to Go, and eventually, AlphaGo was invented
• 07:00 - 07:30 AlphaGo learned by continuously feeding it human game records After learning human game records, it mastered the subtle strategies of Go and began to defeat Lee Sedol After Lee Sedol was defeated, AlphaGo kept evolving, eventually producing a Master version of AlphaGo Then, in 2017, they developed another version called AlphaZero It is said that AlphaZero completely abandoned the human self-training process Using a very simple algorithm, it played games against itself
• 07:30 - 08:00 Whenever it won, it continuously reinforced that algorithm In just a few days, it surpassed the Lee version of AlphaGo In 21 days, it surpassed the Master version of AlphaGo In 40 days, it surpassed all previous versions of AlphaGo in history That version, when compared with the top human Go players, was at least 2 to 3 points weaker The gap between human top-tier players—the most advanced and formidable—and these AlphaGo versions is enormous
• 08:00 - 08:30 At that time, AlphaZero did not play against humans; when it played against Ke Jie, it was still a bit behind After that, the entire field of AI deep learning software began venturing into asymmetrical games Because Go is a game with complete information symmetry—both sides can see everything—which is relatively simple Later, for example, in Texas Hold’em or real-time strategy games, AI algorithms have defeated them This path is relatively simple, why? Because every game has a predetermined objective
• 08:30 - 09:00 That is, you set the objective as winning, and once you win, that is the outcome [This path is relatively simple—why?] Because every game is set with a predetermined goal That is, you set the objective as winning, and once you win, that is the result But most of the problems humans think about actually don’t have a goal of "winning" Just like I said earlier, when you write a poem, there doesn't need to be a goal of winning
• 09:00 - 09:30 For example, when I create a State of Play in China or write a copy, it's certainly not about winning But it requires you to gather information, reason based on that information, and analyze it Then, based on the analysis, you need to come up with a conclusion that is somewhat beyond others This is actually something that those Go programs like AlphaGo cannot do It requires a much deeper AI software to complete this kind of task And this is exactly the work done by OpenAI two years ago
• 09:30 - 10:00 OpenAI was originally a non-profit organization, studying human deep learning models It put all human data into this AI system to learn, compress, and reason Eventually forming a human-computer dialogue model that thinks like humans. After its appearance, it was revolutionary Now it has already iterated to the fourth generation
• 10:00 - 10:30 So, why is deepseek a significant threat to OpenAI? Some have compared OpenAI to the Lee Sedol version of AlphaGo from back then And deepseek is equivalent to the AlphaGo Zero version That is to say, deepseek's algorithm doesn't rely on continuously feeding human data for learning and analysis Instead, it directly inputs human truths and doesn’t need to rely on human knowledge or thought processes to continuously learn and improve
• 10:30 - 11:00 You just need to tell it the rules for writing poems, and that's enough It doesn’t need to learn Su Dongpo’s poems or Li Bai’s poems It can directly write poems on its own—this is what makes it powerful So why is it able to bypass the current US sanctions on chips to China? The essence is that its algorithm is more advanced than OpenAI’s
• 11:00 - 11:30 It doesn't rely on an increase in chip computing power; it’s a breakthrough in the algorithm Now, think about it—this is very similar to the era of AlphaGo, isn't it? Now you can understand the shock this has caused Because you have to understand, over the past few years, the US has had a very important policy to prevent China from surpassing in AI
• 11:30 - 12:00 Because everyone knows that in terms of manufacturing, China has already surpassed almost all Western countries So now, what is the US's advantage? It's still in the forefront of internet enterprises, particularly in the AI industry In AI, there are some "moats," such as chips—Nvidia’s chips Why does the US want to move Nvidia’s chips, and also Taiwan’s chip manufacturing companies to the US?
• 12:00 - 12:30 And implement the most advanced chip embargo on China It’s to prevent Chinese AI companies from using the most advanced chips to create their own AI to compete with the US But think about it, now with the appearance of deepseek It shows that even if the US adopts such a chip embargo policy, China still bypassed the chip embargo and developed its own AI product And in terms of performance, if we don't say surpassing, at least it's almost on par with OpenAI
• 12:30 - 13:00 Doesn’t that mean all these previous sanctions have become ineffective? This is the first aspect The second, more frightening aspect, is values Because in the past, people thought OpenAI was rooted in the open internet environment Rooted in the liberal democracy of Western countries, that’s why it succeeded China has an authoritarian system, and anything that comes out of an authoritarian system must be a castrated version
• 13:00 - 13:30 So, its technology can’t be particularly advanced. But with deepseek, look at it now It does have political censorship, it simply refuses to answer certain questions But in other areas, it is indeed quite advanced So what does this show? Doesn’t it show that an authoritarian system can combine with AI? This could even potentially become a tool that supports China's authoritarian regime
• 13:30 - 14:00 This is also a huge blow to many people The CEO of Wukong Myth, Feng Ji, wrote a post on Sina Weibo He said that deepseek is an innovation that will determine the fate of the nation Why is deepseek considered an innovation of national significance? He mentioned six aspects First, it is powerful—its algorithm is now almost on par with OpenAI's Now, no one else can compare Second, it is cheap—this level was achieved with just over 5 million US dollars
• 14:00 - 14:30 While OpenAI has invested over 500 million US dollars in computational power over the past year—so it's much cheaper, a hundred times cheaper So, as you can see, it’s very cheap Third, it is open source—OpenAI is closed-source, but deepseek is open source, connected to the entire internet, and anyone can download it and use it and improve it together. This is something you can’t compare with OpenAI Fourth, it is free—now all services are free, while OpenAI charges I pay $20 per month for OpenAI, but deepseek is free, and the results are almost the same
• 14:30 - 15:00 Fifth, it is connected to the internet—deepseek directly connects to the internet and can fetch information from the web and turn it into a tool for search and reasoning Sixth, it is local—because deepseek is generated using Chinese it has superior capabilities in processing Chinese information, or thinking and reasoning, compared to OpenAI
• 15:00 - 15:30 Because OpenAI’s main language is English, it doesn't handle Chinese information as well as deepseek does It is said that the chairman of Deepseek, Liang, wrote a response Many people on Zhihu said that this response was actually written by deepseek I saw online that many people requested that deepseek respond in the voice of Chairman Liang to an article by Nvidia
• 15:30 - 16:00 or that it should write a statement in the name of Nvidia's CEO facing deepseek I looked at these statements written by deepseek, and they looked pretty impressive So now it’s very difficult to tell whether articles online are written by humans or by deepseek So, in my opinion, this marks the beginning of a major change in this era I’m not that interested in the competition between China and the US, let them fight it out But I want to talk about this from a larger historical perspective
• 16:00 - 16:30 I want to discuss the deep challenge deepseek presents to the development of human society as a whole I actually mentioned this topic a few days ago during a livestream at the end of the year At that time, someone asked me if China’s economic model and China's modernization model, if it succeeds, would still challenge the universal values of the US I remember I briefly mentioned that China’s model, combining authoritarianism with market economy and big data
• 16:30 - 17:00 if it develops, could pose a challenge to human civilization as a whole At that time, deepseek hadn't appeared yet. After deepseek appeared, I expanded on this topic If you look at the history of human civilization, the earliest civilizations were all authoritarian Including Egyptian civilization, Chinese civilization, and Mesopotamian civilization—only Ancient Greece was relatively special Later, after the Middle Ages, during the Dark Ages
• 17:00 - 17:30 and finally with the Enlightenment movement, which also established the sovereignty of the people Humanity entered into industrial civilization. What was the philosophical foundation of this industrial civilization? The foundation was that God does not exist Nietzsche’s great contribution to philosophy, as everyone knows, is "God is dead" What does "God is dead" mean? In the dark ages, we believed that humanity was merely God’s servant
• 17:30 - 18:00 Above us was a God who was omnipresent and omnipotent The boundaries of human thought could never reach God So, as you can see, authoritarianism aligns with the idea of God That’s why ancient Chinese emperors were called "Son of Heaven"—the emperor’s power was divinely ordained We worshiped them, we accepted their leadership because they were representatives of God God is greater than humans, God is smarter than humans, God is stronger than humans
• 18:00 - 18:30 So we, as humans, must listen to them, and that’s how authoritarianism was built Nietzsche’s philosophy of "God is dead" means that God does not exist, that God is a myth, and every individual is their own master Humans must be free, and the most important foundation of Enlightenment was, in my view, this concept
• 18:30 - 19:00 After God is dead, humans realized that human reason actually has boundaries, and humans are imperfect Each of us has limitations in our thinking, and collective human thinking also has its own boundaries This brings about a problem: humans cannot build a perfect world So, in terms of political systems, we require what? A democratic constitutional system The logic of a democratic constitutional system is that the majority decision solves the legitimacy issue of decision-making
• 19:00 - 19:30 We all vote to choose someone to be president For example, Trump: once elected president, this decision is a legitimate choice Because decisions made by the minority have no legitimacy, but decisions made by the majority do But this majority decision itself doesn’t solve the problem of its correctness So, to solve this issue, the constitutional system established a separation of powers: legislative, judicial, and executive The majority decision only influences the executive system, but it cannot interfere with the legislative or judicial branches
• 19:30 - 20:00 Through checks and balances, human rationality’s limitations and flaws are minimized So, all those democratic philosophers said that democratic and constitutional systems are the best possible systems, meaning they are not perfect, but the least bad If the majority elects an idiot, then four years later, he’s out We can try again. This is the essence of democratic constitutionalism But as you can see, when AI appeared, the rules of the game in this world changed
• 20:00 - 20:30 Humans have created a new kind of God Because in the past, it was unlikely to find a system that surpassed the collective level of human thought and decision-making Our IQs are all quite similar, not too different, but AI is different AI could potentially surpass humans in all areas, not just in chess—chess doesn’t matter
• 20:30 - 21:00 But it could surpass humans in all other fields At that point, we’ll realize that the game rules of the entire society and world we’ve constructed may collapse Why? Because we needed a democratic system because there was no God But if there is a God, why do we need to vote? Why do we need majority decisions? Why not just listen to God?
• 21:00 - 21:30 Why do we need to pursue procedural justice? Procedural justice’s philosophical premise is that we cannot achieve substantive justice Because human reason has boundaries But if AI can surpass all of human reasoning, After an incident occurs, AI can directly determine who the criminal is
• 21:30 - 22:00 Do we still need to conduct trials in court based on evidence? No, we don't We can just listen to AI, right? Or, we can just listen to God Why do we still need to make decisions ourselves in a democratic system? Why not just let AI decide? We could just listen to God, right? So why do we need democratic constitutional systems? This is a profound threat to human society
• 22:00 - 22:30 This isn’t just a problem for China’s authoritarian system Even in the US, with OpenAI, this will eventually be a problem If an entity surpasses humans in all decision-making areas How will we as humans cope? Should we listen to it, or continue making decisions based on a supposedly "least bad" model? Clearly, there’s a better decision, but we still cling to the idea of making decisions ourselves
• 22:30 - 23:00 This is the dilemma I believe that the emergence of OpenAI, or AI in general, could cause the collapse of the social foundations built since the Industrial Revolution In my opinion, this is the most profound challenge for the future world This is the first point The second point is that all human jobs may be fully replaced by AI
• 23:00 - 23:30 For example, after autonomous taxis, the taxi industry would no longer exist For example, the advent of robots means we may not need manual labor anymore But AI’s emergence will lead to a complete replacement of human jobs Many people say, "Look, in the past, all revolutions replaced old jobs, but they always created new ones"
• 23:30 - 24:00 For example, there were carriages before, then cars, and now we have airplanes But the appearance of AI is different; its replacement of human jobs is comprehensive Almost all the jobs we have now could potentially be replaced For example, lawyers—AI’s analysis of cases could surpass the efforts of a lawyer who has painstakingly written a lawsuit in just a few seconds For example, doctors—under the assistance of AI, the diagnosis of diseases could quickly surpass the experience and knowledge from years of medical school For example, even creative roles like poets, painters, and musicians
• 24:00 - 24:30 AI has already started to create music, and while it might not yet match the great masters, there are already cases where AI has learned Mozart's compositions and can create a symphony in Mozart’s style. It can do that now Soon, AI might completely surpass humans in all areas of artistic creation At that point, all human jobs may be replaced by AI, as it could outperform us in everything
• 24:30 - 25:00 This brings two major problems: the first is that if all these jobs are replaced, where will the income for ordinary people come from? We will have no income There may be work for a few AI companies, but everyone else will be replaced So, without income, what will these people do? Didn’t Musk say that in the future, if this happens, the state should give people money to support them, like raising pigs? But I don’t know if this is possible, but it's a problem
• 25:00 - 25:30 The second, deeper problem is that most people think of work only as a tool for survival We work nine-to-five, but in reality, work is also the value of our existence Isn’t it? We work to provide service to other people in society This service itself proves the value of each individual’s existence Humans are the purpose, and one important reason is that we provide services for others
• 25:30 - 26:00 But if we are merely fed by AI, if we are fed by society, then are we still the purpose? Perhaps we are no longer the purpose, and we’ve just become a tool Think about the difference between pigs and humans. The difference is that pigs are born to be raised But in the end, they are slaughtered. If they weren't slaughtered, wouldn’t they also just be tools?
• 26:00 - 26:30 So if one day we no longer need to work, and we just do nothing, unable to do anything better than AI, what purpose does our existence have? Isn’t that a deep existential crisis? In my view, this is the second aspect The third aspect is the potential rise of superhumans
• 26:30 - 27:00 As everyone knows, Musk is researching brain-machine interfaces The brain-machine interface is about combining AI’s computing methods with the human brain Integrating it into the human brain so that people can use AI’s intellectual power to enhance their thinking levels So, is it possible that superhumans could emerge? If Musk had such a tool controlling his brain He could possibly surpass the IQ of every other human being, which is one aspect
• 27:00 - 27:30 Another aspect: a few years ago, China had a case with He Jiankui, who was doing gene editing He edited genes to prevent a child from being infected with the HIV virus He knocked out some genes, and during the child’s birth process, the genes were edited This caused a shock in the international scientific community Because this was the first time humans edited their own genes
• 27:30 - 28:00 After the incident, He Jiankui was sentenced to three years in prison, but he has already been released But this may be irreversible from a technological standpoint In the future, the entire human genome will slowly be decoded The first phase may be the promotion of genetic editing techniques for correcting deficiencies For example, high blood pressure and diabetes in humans could be caused by certain genetic defects Genetic modification might be able to prevent these diseases, so you could drink soda and never get diabetes
• 28:00 - 28:30 You could eat fatty foods and never get high blood pressure. This might be acceptable to many people Then, human virtues, such as courage, beauty, and strength, might also be caused by certain genes For example, traits like courage, beauty, and strength These qualities might also be caused by certain genes So, if we modify our genes appropriately, you could acquire these virtues—would you do that?
• 28:30 - 29:00 Once this becomes possible, it opens up a Pandora’s box. Why? Think about it—there’s a principle of fairness in human evolution In the process of generational evolution, the rules are relatively fair For example, in this generation, you worked hard and became someone like Jack Ma But Jack Ma's children might not be as smart as him Or even Elon Musk, you have so much wealth, but your children may not listen to you and change their gender
• 29:00 - 29:30 Maybe in two or three generations, Musk’s descendants will become ordinary people Meanwhile, somewhere in a slum, a new Musk is born through God’s dice roll This person struggles and becomes the next world’s wealthiest This is the fairness principle of God’s dice roll in the gene evolution process, rolling the dice for each generation Although some rich people can use their resources to improve their children’s opportunities
• 29:30 - 30:00 If you’re born in Bill Gates’ family, you can access the best education, the best resources But you can’t make up for genetic predispositions However, if genetic editing becomes practically possible Then the wealthy can ensure their children gain an absolute genetic advantage that ensures their success in future generations
• 30:00 - 30:30 Elon Musk could design a son who is not only smart but also brave, beautiful, and charismatic And his IQ might have a gap that’s as wide as an ocean compared to others So will there still be equality in this world? It’s likely that there won’t be any equality So, how will we face this world in the future? What challenges will these new rules pose to the entire human society?
• 30:30 - 31:00 So, after deepseek’s appearance, I personally believe the so-called AI arms race between China and the US is a small issue It’s just a minor ripple in the grand course of history The real crisis is the impact artificial intelligence will have on the entire course of human civilization This is something many people haven’t fully considered The scenario I’m describing may occur in the next 20 years—the real future world has already arrived
• 31:00 - 31:30 The future world, which we cannot even imagine, is already here It may be something like 1984, or it could bring an unprecedented shock to human society I don’t know what that future is like From a certain perspective, I’m quite pessimistic about the direction of human civilization and progress What I’m saying today is really to share with you my own sense of melancholy about the future in this new uncertain era
• 31:30 - 32:00 But one thing I can confirm is That in facing this era, we must constantly learn and constantly surpass ourselves Otherwise, the first to be eliminated will be you, not others This is likely the most valuable thing for each of us Alright, that’s all for today, thank you everyone