How to solve the problem of Artificial General Intelligence (AGI). Part - VI : Universal AI

Universal AI tries to solve the problem of goal driven AI where we need to define correct environment for an agent to operate. Using the universal theory of induction proper sequence of decision could be made.

Algorithmic information theory deals with Kolmogorov complexity of algorithms. The minimum length of an algorithm to solve a problem is also related to the concept of entropy. In the case of Artificial General Intelligence measuring Kolmogorov complexity is important because it will generalize the problem efficiently.

Agents operating in probabilistic environments is a common AGI context. Universal AI could solve this problem if applied heuristically.

We could use Markov decision process to define optimal policies. It could also be used in methods of reinforcement learning. To solve the AGI problem the issue of immortality of an AI agent need to be solved also.