Wednesday , January 27 2021

Ilon Mask, for which question I would like to ask super-advanced artificial intelligence

Tesla and SpaceX chief Illon Mask said in an interview with Massachusetts Institute of Technology researcher Lex Friedman that his first question about super-advanced artificial intelligence would be related to simulation theory. During a very peaceful conversation about the future of drones, it started within the podcast, the theme flowed smoothly first in the field of physics, then love, and then approached the question of whether our Universe is someone's computer simulation.

"Perhaps you, and perhaps someone else, will once create a very powerful system of general artificial intelligence (AIL) and you will have the opportunity to ask one question. What kind of question will this be?" Friedman asked until the end of the 33-minute announcement .

The mask paused for a few seconds, lowered his eyes to the floor and began to drive them very quickly by the side, obviously, very carefully choosing the right answer. Finally, he said:

"What's out of the simulation?"

Musk shares his idea that humanity lives in a very large computer simulation as early as 2016, noting that given all the strangeness that is happening in our world, the probability of this is about "one million". As an analogue resolution, he cites how fast and very sharply the video games are developing. After all, just a few decades ago, the whole world was "punctured" to Pong – one of the first video games that represented a tennis simulator, where a small square acted like a ball and two rectangles like a racket. It took us only a few years to learn how to make photorealistic simulations of fictional worlds, so why not suppose someone from the outside created an even more advanced photorealistic simulation in which we all live?

"Let's imagine that we have been transported for 10,000 years in the future, which is nothing in terms of evolutionary proportions. The development of video games is already beginning to approach the full photorealism, when the computer-generated image becomes more difficult to distinguish than reality Now imagine that for many centuries the level of technology will be a thousand times higher than today and that the most realistic games can be put on billions of consoles, computers, or any other toy device. In view of all of this, we can safely assume that ne of these billions common games is our life today. "

The initial interview with Mask was about how much more advanced unmanned vehicles of the future, as well as how to use the neural network that collects data from eight installed cameras and 12 ultrasonic sensors in each Tesla car, will enable the production of fully unmanned vehicles. In the end, the conversation turned to how these nerves could serve as the basis for creating a general or universal artificial intelligence at the human level.

"This technology will appear very quickly, then we need to decide what to do next. Of course, if we have even such an opportunity," Musk said.

The head of Tesla and SpaceX in the past also founded Neuralink, whose goal is to create a universal computer interface that allows you to combine the human brain with a computer. In addition, Ilon supported the creation of a set of rules for the development of AI, noting that without control of this sphere, sooner or later, we will all be in great trouble.

During the interview, Musk also noted that if there is no physically determined distinction between reality and simulation, then it is very likely that testing that our reality is a simulation will also be impossible.

"There may be ways to test whether everything is around is a simulation. You can, for example, assume that the computer that controls it will determine that the subjects found in it have found the right way to check if the world around them is a simulation. in this case start the error correction program; the simulation can pause, and maybe even completely – start again, "commented Musk.

Some experts in the field of artificial intelligence are confident that the super advanced general artificial intelligence will emerge in the next ten years. Perhaps, Maska will still be able to personally ask the system for an issue of interest.

You can chat about the article in our conversation with Telegram.

Source link