Artificial intelligence (AI) algorithms have had some success in learning the laws of physics. In 2009, researchers at Cornell University fed data from a swinging pendulum to their AI, and it managed to learn laws such as the conservation of momentum and Newton’s second law of motion.
In 2018, researchers at MIT went further, showing their ‘AI physicist’ a bouncing ball in various simulated worlds, which allowed it to derive the laws for each simulation. Like a scientist, the algorithm could even combine individual laws to make unified theories.
To discover E=mc2, we would need to show an AI object behaving according to Special Relativity. For example, we could create a simulated world in which the mass of an object appears to increase with speed. Let an AI loose in this world and it may well figure out Einstein’s famous equation linking mass and energy.
Read more:
- When we start uploading our brains to computers, will our sense of self be uploaded too?
- Does a computer know that it’s a computer?
- What is intelligence?
- If we’re ever able to make robots as intelligent as us, won’t forcing them to work for us be as bad as slavery?
Authors
Luis trained as a zoologist, but now works as a science and technology educator. In his spare time he builds 3D-printed robots, in the hope that he will be spared when the revolution inevitably comes.
