RANCHO PALOS VERDES, Calif.– In a far-ranging conversation at the Code Conference, Tesla and SpaceX CEO Elon Musk discussed his view on self-driving cars, sending spaceships to Mars, and why he thinks we're probably living in a simulation.
That last and most surreal point came toward the end of the conversation, in response to an audience question. Musk talked about how games have progressed from Pong to photo-realistic 3D games that millions are playing, and how we are beginning to see virtual reality games and simulations. If you assume any rate of improvement at all, games will become indistinguishable from reality at some point, certainly within the next 10,000 years, Musk said. According to him, the odds that we're in a "base reality" are one in billions, and the two most likely scenarios are a simulation indistinguishable from reality or that civilization will cease to exist. This is a conversation he said he's had so many times, he has now banned it from being discussed when he's in a hot tub.
The conversation, with conference hosts Walt Mossberg and Kara Swisher, began with a discussion of SpaceX, which recently was able to land an orbital stage rocket on a ship in the ocean. Musk explained the differences between sending a ship into space and into orbit, and why it is so important to reuse the booster stage of the rocket, which he said accounted for about 70 percent of the cost, or about $35 million.
Among more immediate plans, he said SpaceX plans to re-fly one of the landed rocket boosters in three to four months; and by the end of the year, he hopes it can launch the Falcon Heavy, which is designed to provide 5 million pounds of thrust. This will mostly be used for satellite launches. Next year, he hopes to launch the first Dragon version 2, which he said would be able to ferry up to seven astronauts to the space station and be used as a general science delivery system anywhere in the solar system.
Musk said he will be presenting an architecture for Mars colonization at the IAC conference in March, with the goal of transporting large numbers of people and millions of tons of cargo to Mars, allowing eventually for a self-sustaining and growing city in Mars. In the meantime, he said, SpaceX will send a mission to Mars starting in 2018. Initially, this will just be cargo, but he said the company should be able to launch people in 2024 with arrival in 2025.
Like Jeff Bezos earlier in the conference, Musk rejected the concept of a "Plan B" for abandoning the Earth. Musk said he believed it is important for us to ultimately become a "multi-planet species," having life as we know it expand to the rest of solar system and eventually to other star systems. We need things that are inspiring, he said.
On self-driving and electric vehicles, he joked that there have been so many announcements, "I'm waiting for my mom to announce one." But overall, he said it was "encouraging to see all this activity" and that this was good for the industry.
Tesla's goal is to get to half million cars total by 2018 and a million cars a year by 2020, and that he can see a clear path to get there. For the Model 3, the company has booked around 400,000 consumer orders, which he said "caught us by surprise." The goal there is to make a great but affordable car, starting at $35,000, when the average cost for gasoline cars is about $32,000 in the US and the electric vehicle requires no gasoline and less maintenance.
"I don't think any of the car companies have made a really great electric car yet," he said of traditional car companies.
On the difficulties in making cars, "the sheer scale of automotive manufacturing is hard to appreciate," he said. Not only does Tesla have a huge assembly plan, it has a complex supply chain.
He noted that the "Gigafactory," where the company is producing batteries, will be the largest footprint building in the world when it is completed. There is a lot of misunderstanding about the different kinds of lithium-ion batteries; "we don't have a range issue," he insisted, since the Model S can go 300 miles. We could have a 400-mile range car today, he said, but what really matters is reducing the cost per unit of energy.
Musk suggested it will be odd to have a car without autonomy in the future. Google isn't likely to be a direct competitor to Tesla, but will be more likely to license its technology to other car companies, according to Musk. But he expects more direct competition from Apple. He noted that the car business is fragmented with a dozen significant car companies, none with more than about a 10 percent market share.
"Autonomous driving is a solved problem," Musk said, explaining that slow-speed urban driving and highway driving without barriers are easy; it's difficult to drive 30-40 miles per hour in an urban environment. We are less than two years away from complete autonomy, but regulatory approval will take at least another year.
Regarding Tesla's plans for a fully autonomous car, the company will have another event around the end of the year to talk more about it. He wouldn't give details, but said "we're going to do the obvious thing."
On the advantages and threats of artificial intelligence, where Musk has been outspoken about talking about the dangers, he said his full position would require quite a long explanation, but that he was concerned about certain directions AI could take that wouldn't be good for the future. "Not all AI futures are benign," he said.
He said he was particularly interested in democratizing AI power, and that's why he helped found Open.AI. He said this was a non-profit, but unlike most, would be developing technology at a fast pace with a high sense of urgency. He said he was worried about "the idea of living under a despot," whether a computer or the people controlling the computer; and worried that a single AI could result in a huge concentration of power. Asked which competitor he was talking about he said "I won't name a name, but there's only one." But he said Open.AI isn't about directly competing, but more about trying to increase the probability that the future would be good.
Asked by Mossberg if he worried that by making the AI open, it might get used by some bad actors, Musk talked about how if AI is widely distributed and we can link AI power to each individual's world, if somebody tried to do something terrible, the collective world of others could counteract it.
Musk said that if you think the advancements in AI have been "quite astonishing" and you assume any rate of advancement in AI, people will be left behind by a lot. "We would be like a housecat," he said, and that's the benign outcome. One solution he suggested might be a "Neural lace" interface to connect an AI to your cortical neurons using your veins and arteries, providing you a high-bandwidth neural interface with your digital self.
This would be an alternative to god-like AIs, he said, and though he isn't saying he would do this, "somebody's got to do it," he said.
On the Hyperloop project, he said he had an initial idea, but that originally he got it wrong. But, he said, with a lot of iteration he came up with something where the physics hangs together.
He said he thought it would be good to have safer, less costly, more convenient new transportation systems, and that the Hyperloop would be "better than a fast train that isn't as fast as what the Japanese built in the 80s."
But he said his plate is full running Tesla and SpaceX, so he just wanted to encourage anyone who is interested and give them support, but that he is not an investor in any of the companies working on it. If companies trying to do it now don't succeed, he said he might do something in the future, but said "I don't want to front-run them."
On other topics, he said he worried that there were a few too many talented entrepreneurs in the Internet space, and it might be better to have some of that talent in other industries.
Latest Breaking Technology News and Electronics News Headlines & more