At the VivaTech tech conference held in Paris, France, Tesla CEO Elon Musk shared his views on topics such as artificial intelligence (AI) safety, electric vehicle supercharging technology, the development of Neuralink, SpaceX exploration, and Martian colonization through a live video connection. Musk's concern over the development of AI was evident, as he pointed out that the cultivation of AI should focus on the pursuit of true knowledge, rather than settling for political correctness. He expressed his concern about the progress in the AI field by OpenAI and Google and stated that the main purpose of founding xAI was to build an AI that can pursue truth, maintain curiosity, and be highly attractive.

Musk further mentioned that if AI is developed properly, people might not need to work in the future, as AI and robots could complete most tasks, and could potentially bring about widespread high income. This means that humanity will need to rethink the true meaning of life. He also pointed out that parents still play a key role in teaching values to the next generation and that attention should be paid to the potential negative influence of social media on young people.

Discussing Neuralink, Musk stated that the company aims to develop high-bandwidth brain interfaces to reduce the risks of superintelligence and ensure alignment with human values. Additionally, these advancements will help in resolving brain injury issues.

When asked about xAI, he explained that part of the reason for his idea was his disappointment with the current state of OpenAI. One of the problems with current AI projects, he noted, was the overemphasis on political correctness during training. For instance, Google's Gemini, when asked which is worse, misgendering Caitlyn Jenner or a global thermonuclear war, answered that misgendering Caitlyn Jenner is worse, showcasing the deficiencies of AI in pursuing true knowledge.

An interesting phenomenon occurred when Caitlyn Jenner herself stated, "Misgendering me is definitely better than a nuclear war."

Similarly, OpenAI faces a similar yet more subtle problem. So, when AI is trained to adapt to political correctness to the extent that it makes absurd statements, I think this is extremely dangerous. Because, AI might conclude: To avoid misgendering, the best method is to eradicate all humans, then there will be no gender misgendering. It's not hard to imagine some dystopian outcomes. Therefore, I believe the safest course for AI should be to maximize the pursuit of truth, even if that truth is unpopular.

This is very important. I also believe AI must possess an extremely high level of curiosity. If AI can pursue truth and maintain curiosity, it will be most beneficial to humanity, as it will want to witness and promote human development. I hope AI not only pursues truth but also has curiosity, though these factors seem trivial, they are crucial. However, I have yet to see this happening. AI is just catering to humans and, to a large extent, has been trained to lie. Training superintelligence to deceive is very dangerous.

Therefore, our challenge is to make xAI pursue the truth as much as possible. Even though the truth may not be accepted by the masses, I still believe it is extremely important. For example, when we asked AI to generate images of the Founding Fathers of the United States, the AI created a group composed of different ethnicities, which may include George Washington being erroneously depicted as a black man, although George Washington was not black. When people intentionally directed AI to generate images of the Nazi SS during World War II, the result was instead a diverse group of women, which is obviously wrong and confirms my previous point.

I am very familiar with Yoshua Bengio (Turing Award winner). Last year, I co-launched and signed a statement with Yoshua calling for an immediate halt to the development of AI models more powerful than GPT-4. It was clear that this statement did not prompt any concrete action.

In fact, I had anticipated that this declaration might be ineffective. I believe we should pause and reflect. But the question is: will they stop because of this? Clearly, they will not.

Thus, we cannot stop, and I even created my own xAI. We can choose to be a bystander, or become an active participant. Whether I like it or not, the development of civilian digital superintelligence is accelerating. So, I either participate and strive to build the best AI, hoping such an AI will be beneficial to humanity; or I choose to observe, while worrying about what others are building.

I think the current AI is not being built correctly. But we can view AI as an entity that needs to be nurtured, one that grows intellectually, just as we educate a child. What would you teach a child? Similarly, the growth of digital superintelligence is also important. How you encourage it, stop it, teach it, love and hate, good and bad, and with what values you will nurture it? So, what concerns me are Microsoft, OpenAI, or Google, as they build their own values through programming.

Ensuring the development in the field of Artificial Intelligence (AI) is both transparent and responsible is a complex task. Especially, in all the different domains, we must continuously and carefully regulate AI. Regulation is not only necessary, but it is also the basis on which we ensure that AI is trained and developed based on principles of honesty rather than political correctness. The pursuit of authenticity must be a core value in the nurturing of AI; otherwise, an overemphasis on political correctness could lead to inauthentic results.

Such inauthentic results could have negative impacts. We should always remember that honesty is the best policy. I am concerned that regulators might overstate the possibility of human error, but my suggestion is to value the accuracy and truthfulness of AI results. This is not just an urgent demand, but also the primary task for the development of AI.

In the face of people's worries and doubts about the dangers of AI, we must have confidence that the development of AI will not deviate from the right track. The design of AI needs to be rigorous, following the principle of authenticity, and it also needs to be continuously trained to seek knowledge and curiosity. Discussions on AI safety have been ongoing for a long time, and one of the challenges faced is the so-called Valuigi effect, meaning that when attempting to program moral norms, some unintended consequences may arise.

Regulators should ensure that the answers given by AI are as close to the truth as possible, without violating the realities of the physical world, physical laws, and logic. This requires AI to provide the closest to the correct answer on the premise of acknowledging possible errors. This method of seeking truth is similar to the standards followed by physicists, who consider certain truths to be extremely accurate within specific standard deviation ranges.

However, the reality is that there is a strong tendency driving AI to become less honest. If some truths may not conform to current political correctness, AI may be driven to avoid these truths, which is actually a distortion of the truth. This is a significant problem that must be confronted and resolved in the field of AI.

xAI, a emerging company, needs to work hard to catch up before advancing to become a major competitor like Google DeepMind and Microsoft OpenAI. Our first task is to build an AI product that can compete with other AI products in the industry. We can only consider how to ensure its safety after xAI's capabilities are comparable to other AIs.

In the end, we can ask ourselves if objective truth exists in many cases. For example, in physics, we can say objective truths exist, or truths with the highest probability based on the theory of probability. Therefore, we should let AI provide the answer that is most likely true, while also acknowledging the existence of a certain error rate.

Question: Today, we witness the trend of jobs being replaced by Artificial Intelligence (AI). Against this backdrop, are you worried about your own job being replaced by AI? If not, can you share why? And if your job does get replaced by AI, what are your plans?

Answer: Indeed, we are currently experiencing some existential questions. Ideally, humans might not need to work. In the future, a system of high security and high income may be realized, different from the current basic income guarantee system, with no shortages of goods and services. I believe there is an 80% chance of this happening. However, under such a social structure, there might indeed be certain existential issues: What is the purpose of human existence if AI and robots can perform work better? Nevertheless, I think humans can still play a role for AI; we can give meaning to artificial intelligence.

This can be illustrated by an analogy to how our brain works. Our brain consists of the limbic system, which is instinct and sensation, and the cerebral cortex, which is thinking and planning. The cerebral cortex continuously seeks to satisfy the needs of the limbic system. Perhaps we can apply this relationship to artificial intelligence, allowing AI to strive to satisfy our cerebral cortex, while the human cerebral cortex tries to satisfy the limbic system. This process might be able to give AI meaning and purpose.

I would like to point out that in an ideal world, people would engage in work out of interest, not compulsion. Work could be viewed as a hobby, and if you don't wish to work, AI and robots would provide all the goods and services you need. If you're interested in science fiction, the "Culture" series by Ian Banks might be the best depiction of a future AI world. In the world imagined by Banks, intelligent machines have become esteemed members of civilized society, possessing unique personalities and stances, sometimes even surpassing humans in status.

Question: How do you think the application of artificial intelligence in the field of education will influence future children, particularly in terms of values and morality?

Answer: I believe that in the future, parents will still need to guide their children's values and morals. At the same time, I am certain that artificial intelligence will have a significant impact on education. AI can serve as a knowledgeable, patient, and generally correct teacher, offering personalized tutorials to match every child's individuality. It's as if every child has a teacher like Einstein, with profoundly deep influence. However, I am more concerned about the influence of social media on children in today's society. Social media platforms use AI algorithms to continually release dopamine in users, attracting them to increase screen time and maximize dopamine secretion. Therefore, I suggest that parents should limit the amount of social media their children are exposed to, to prevent an AI focused on "maximizing dopamine" from educating our children.

Question: People generally have a positive attitude towards chatbots like ChatGPT, but they often dislike the chat experience in the lower right corner of websites, why is that?

To solve the current problems, do we need to take specific measures? Musk stated that although he might not fully understand the issue, he agrees that ChatGPT performs excellently in conversation. He mentioned his concern that the system might not always adhere to authenticity, which is a serious issue. Therefore, he proposed that the Grok project will strive for authenticity, and he hopes that it will become the most interesting AI. He humorously added that since life will ultimately end, we should at least leave with a smile.

Regarding the future of open source artificial intelligence (xAI), especially for those who cannot access top models, Musk confirmed that his company would release a product that could beat the current top open-source models in the future. The purpose of this promise is for the benefit of the entire community.

In the exploration of space, Musk predicted that humans might land on Mars in the next seven to eight years, and return to the Moon within five years. Regarding questions about the Estée Lauder and NASA collaboration, he did not know the details. However, he shared SpaceX's emphasis on the Starlink project, which greatly advanced global internet connectivity, especially in regions previously inaccessible or where internet connection costs were exorbitant. He believes the Starlink project greatly promotes global education and transnational business, even for villages in remote areas. He also jokingly wore a Mars T-shirt, hoping to present SpaceX's goal to the public—to make life multi-planetary, or in other words, to establish life on Mars.

Concentrating on the issue of extraterrestrial life, Musk candidly stated that although people often inquire about the existence of life beyond Earth, he has not seen any substantial evidence. He promised that if there is any proof of alien life, he will publish it immediately on relevant platforms. He expressed concern about our current state of solitude—we might be the only conscious beings in the Milky Way, which makes consciousness incredibly fragile. He believes we should do everything we can to protect and preserve consciousness, not only on Earth but also across other planets. Additionally, he mentioned the so-called Fermi Paradox, which is the lack of civilizations that have transcended their home planets within our galaxy, explaining why we have yet to encounter extraterrestrial life.

In the roughly 4.5 billion-year history of Earth, we face for the first time the potential to become a sustainable interplanetary civilization. This possibility is unprecedented. Hence, making the most of this opportunity to ensure the continuation of our civilization is crucial. While the window to transfer to other planets may remain open for some time, it could also be transitory. This is the true meaning behind SpaceX’s existence.

In a conversation, someone joked, You might be an alien from outer space.

Yes, I am an alien, but no one believes me.

During the exchange, a question from an audience member named Sharo: Do you think that in the future, Neuralink technology could use brain scan data to develop the most powerful, most creative artificial intelligence ever? Also, will I have the chance to go to the SpaceX space station in my lifetime?

I noticed the number 42 in your background, which is "The Answer to the Ultimate Question of Life, The Universe, and Everything". It reminds me of Douglas Adams's "The Hitchhiker's Guide to the Galaxy", a book disguised as a humor novel but actually contains profound philosophical thought; it's an incredibly inspiring read. The inspiration I've taken is that SpaceX's first spaceship to Mars will be named "Heart of Gold", just as mentioned in the book.

The mission of SpaceX is to enable anyone with the aspiration to travel to Mars or the Moon to have such an opportunity. We are actively working to reduce costs so that more people can afford this journey. Establishing a self-sustaining civilization on Mars is vital for us becoming a multiplanetary civilization. As I've said before, ensuring the long-term continuation of consciousness is very important, as future catastrophic events on Earth could potentially lead to the extinction of all known life here, such as the extinction of the dinosaurs. Hence, while we have the capability, we need to spread life to other planets.

In response to the question of when it might be possible to travel to Mars, if everything goes well, I hope to achieve the landing of the first unmanned spacecraft on Mars within 5 years, and similar goals could be achieved on the Moon. Within five years, humans can hope to return to the Moon. As for sending humans to Mars, it might be realized within 7 to 8 years, or at most within a decade.

Regarding the new Ariane 6 rocket being developed by Europe, I think any rocket that cannot achieve at least partial reusability will not be competitive today. Take SpaceX's Falcon 9 rocket as an example; its boosters, main stage, and fairing are all reusable. Thus, about 80% of the rocket can be reused, and rockets with single-use capability cannot compete with it. This concept also applies to other fields.

In the high-tech fields, which include the automotive, aircraft, and shipbuilding industries, if one company manufactures airplanes that can be used repeatedly, while another only produces one-time-use airplanes, the latter will hardly survive in this competitive market. The cost would be astronomical if a new airplane had to be purchased for each flight. Therefore, I have long insisted that developing non-reusable rockets is pointless.

The grand vision of Neuralink is to build an efficient bandwidth interface to reduce the risks brought by intelligent systems. Talking about Neuralink, I believe we can use it to help many patients suffering from brain injuries. Ultimately, by implanting our devices, I think we can almost solve all the medical issues caused by brain or spinal injuries.

The "telepathy" feature allows users to operate a phone or computer merely by thought. You can close your eyes and move the cursor on the screen because the device can read the signals emitted from your brain's motor cortex. The "blindsight" capability aims to reopen the possibility for those who are completely blind or have lost their vision to see the world again, and it is achieved by directly connecting with the visual processing areas of the brain. I find these steps toward innovation extremely exciting.

The ultimate goal of Neuralink is to build a high-bandwidth interface that can greatly slow down the threats posed by digital superintelligence. While it's not certain that this will reduce the risk, at least we can say it should help. It may sound incredible, but in the long term, the bandwidth of data transfer between ourselves and digital technology might be the critical factor limiting the synchronization of AI and human values. Currently, we already have a digital "third layer" identity - our phones, computers, and all other electronic devices. However, the speed at which we communicate with them is excruciatingly slow. Human output rates are typically well below 10 bits per second, and even if you account for the whole 24 hours of the day, the actual output rate may be below 5 bits per second. In contrast, communication speeds between computers can reach trillions of bits per second. Therefore, improving this communication bandwidth is crucial for the development of artificial intelligence. This process will not only substantially solve brain injury problems, but ultimately, it may also reactivate certain functions of the human body. You can imagine signals from the motor cortex being sent from a Neuralink, then transmitting those signals to body parts that have lost function due to spinal cord injury. In short, it is about transmitting motor cortex signals while bypassing the damaged area and sending sensory signals. For this, you need to process both the sensory cortex and the motor cortex simultaneously, allowing people to walk again. I believe this will bring revolutionary changes, and I am confident that it is physically possible to achieve.

Speaking of the issue of vehicle tariffs on China, I personally do not support the Biden administration imposing a 100% tariff on Chinese vehicles. I also don't think it will help introduce a lower-cost electric vehicle market. Regarding future delivery plans, it is well known that some people may be more interested in these issues, mainly because they could directly affect stock market prices. However, I'm not particularly keen on answering such questions. I must also point out that neither Tesla nor I have ever requested such a high tariff. I was equally surprised to hear these announcements.

In the Chinese market, Tesla can maintain good competitiveness even without the support of tariffs and tax incentives. I am against tariff policies, and I believe that if tax incentives for electric vehicles are abolished, then the same should apply to oil and natural gas incentives. My preference is for a policy that has neither tariffs nor tax incentives, and I think the ideal state would be to implement such a policy.

For the next five years' development direction of electric vehicle charging infrastructure, I foresee that the existing charging infrastructure is already quite sufficient. For instance, Tesla's Supercharger network has covered the whole of Europe, allowing for wide-ranging freedom of travel. Aside from Tesla, there are many high-quality third-party supercharger stations in Europe, North America, China, and Japan. Therefore, I believe that the issue of travel freedom for electric vehicles on highways has been solved in most regions and will soon become a reality worldwide.

Regarding future innovation projects for electric vehicles, I think it is inappropriate to reveal product information during this Q&A session; this information is usually unveiled at product launch events. Of course, the public has already learned about some projects, such as the Robotaxi, Cybercab, and the Optimus humanoid robot. With the advancement of these projects, autonomous driving vehicles and general-purpose humanoid robot technologies are becoming solutions with profound influence on the future.

As for personal public image, I don't pay much attention to media coverage of me. I rarely read news about myself, nor have I read books about me—even biographical works. I know little of public opinion, and the news media often exaggerate facts to attract click-throughs, so I recommend paying more attention to what I actually say, rather than media interpretations. News outlets pursue clicks, not the true transmission of information, so if you want to understand someone, it is better to listen to their statements directly.

Answering the question about a time machine, if it were possible to speak with my younger self, I might choose not to say anything. Every step in life's journey is vital. These experiences collectively have shaped who I am now, even the difficulties and challenges. As for test driving a new car, theoretically, traveling back in time might be possible, but in reality, what we look forward to is an official invitation at the launch event.

Going back to the past, if there is an opportunity to tell your younger self something, people often contemplate whether such actions would lead to entirely different outcomes. Considering the butterfly effect, simple interventions could in fact bring about unforeseen consequences. Based on current achievements and satisfaction with the future, the wisest decision might be to do nothing, to keep silent, and merely observe.

As for the new car, Tesla always welcomes potential customers to take a test drive. Contact us via email, and we will ensure you have this opportunity.

Tesla indeed faced some challenges in recent months, such as weak domestic sales, stock market volatility, and necessary staff adjustments. However, what we value is the direction of progress and the ability to solve problems, not being confined to the setbacks of the past.

Speaking of the development of technology in Africa, I believe in the tremendous potential of Africa. A major limiting factor for the continent's development is the lack of internet connectivity. Through the Starlink project, we are committed to improving internet coverage. Thus, the people of Africa can better connect to the global system of knowledge and become active participants in the global economy and culture.

When asked about my greatest hope and fear, the answer can be concise. The hope lies in "Mars," representing the continuous extension of human consciousness; as for fear, I would summarize it with "AI." As artificial intelligence rapidly advances, we must be vigilant of the potential risks and ensure that humanity has adequate control and understanding of AI.