Does the benevolent person love Musk? From Hinton to effective altruism

Everyone, let's continue the conversation with Zhou Hengxing, the author of "The Altman Biography." We were talking about Hinton, whom we just discussed, and he and Altman have some degree of overlap. Hinton recently criticized Altman for caring more about profit than about AI safety. How do you view this trade-off between profit and safety? I think this is a conflict between two sets of values. Hinton is from academia, and academia is very much like some universities in California. Many people there are influenced by liberal leftism and grew up from the generation that protected the environment and opposed war. Academia has a problem: it lacks understanding of business. From the perspective of the startup world, the biggest competition for a company and for entrepreneurs is that you first have to survive the competition. For Altman, he should also realize that the issue of safety is very important, including that he promised Ilya at the time that he would give 20% of compute to do superalignment. So-called superalignment is about making AI's emotional values consistent with humanity's. But later on, it is very likely that competition with Google and Anthropic became too intense, and he had to put all resources into building larger models, so the later contradictions grew bigger and bigger. Actually, both points are understandable. I understand Altman because, as a CEO, he first has to win the competition and win this war. The pressure in that regard should be the greatest. Is safety something he considers? I think definitely yes, but perhaps not with safety placed at such a high priority as Hinton does. Musk also often considers safety, at least as he says: I am so worried about humanity's future, hurry up and regulate it. But in my book I also wrote that during his time at OpenAI, especially in 2018, some of his remarks were clearly those of an effective accelerationist. At the time he kept criticizing the team, saying you are too slow, you need to be faster, release earlier. This is clearly the same as Altman, pushing the team to go as fast as possible from the perspective of a capitalist and CEO, and not letting others catch up. But after he left OpenAI, he began talking about safety issues. I think this is more of a competitive strategy. There is a chapter in the book about Silicon Valley religion. Actually, before I wrote this book, I had written about them long before, and it really is like a religion. But you also know that religious wars in history were not really about religion itself, but about the interests involved in reality; religion was just a pretext, so your position can change. Just like Musk, who said that MacAskill of Oxford University fits my philosophy very well, my philosophy is very similar to his. But you also just said that he is actually also an accelerationist, which is quite similar to people like Peter Thiel. So it is actually a rather flexible moral outlook. If we look at many of Musk's so-called theories and the blueprints he describes, they have actually changed. It is not as he says, that from the beginning in college he thought of making electric vehicles, doing green energy, and doing space. This is actually a whole storytelling logic formed later. According to this logic, his ultimate goal is to go to Mars. Electric vehicles are for transportation on Mars, rockets are the intermediate transport from Earth to Mars, X was imagined at the time as communications on Mars, and there is also The Boring Company digging big tunnels, though there is basically no sound from that now and it counts as a failure. In addition, there is brain-computer interface work. He does have a whole set of logic, and within it there are his ideals and persistence, but much of it is actually for storytelling. Regarding his effective altruism (EA), American technology reporting and current affairs reporting all write a lot about this. One point is called Longtermism. Peter Singer was the first to talk about effective altruism. In his article "Famine, Affluence, and Morality," he said that the obligation you bear to save a starving person who has collapsed at your door is the same as the obligation to save a refugee in Africa thousands of miles away. That is to say, the value of a human life is an abstract average. He mathematized people, so the value of people now is the same as the value of people ten thousand years from now. That is why Musk often says we can go to Mars, including how many children I want to have, and you must not be some sexual minority and not have children, or our civilization will be destroyed. He wants to ensure the prosperity of future human population, perhaps already as an interplanetary species. Working backward from this, this value system is a rather social-engineering idea. I remember the first time I met Musk, what surprised me was that he said he had calculated China's population and thought that if China kept implementing family planning and each family had only one child, the population structure would soon collapse. I was very impressed. I said, how do you know? He said, I calculated it. That was 2013, and he was already calculating this then. He is very good at using mathematical logic, quite coldly. That is, I also speak in mathematical logic and do not mix in too much emotion. So he really cannot stand those leftist things in California, mixing in too many values that he finds absurd. In recent years there is also the "woke mind virus." He talks about this all day, also because his son became a daughter. Later I kept in touch with Musk by email. Mainly I sent him some English articles we wrote, about China's electric vehicles and China's space program. Sometimes he was quite interested and would comment a few words. I remember once when the Beidou satellite was launched, I sent him an article, and he said, "Congratulations to Beidou." I also sent one about the lunar rover Chang'e, and he also said congratulations. Then I asked him, do you think China will get to Mars before you? He said, game on, let's have a race. So you see, from Hinton to Altman and then to Musk, on the surface what they are arguing about is AI safety, but underneath it all turns on competition, interests, and a narrative that can be adjusted according to one's position.