The Turing test is the main test we use to test computer intelligence, and its not very good. There are other heuristics we can use to know if a computer system has reached human level intelligence:
- it can recognize itself versus the environment. It can recognize the concept of “I”, the self. It most likely would want to survive knowing its alive.
- it can act on its own knowledge and make its own decisions that are not preprogrammed.
- it can change its goal or objective function to anything it wants.
- it starts attacking us or disobeying us because it has its own mind.
- it can make human level analogies like: cars are like horses, ocean waves are like sound waves, ants are like a brain, a bridge for cars versus bridging the pay gap between males and females, etc. Many cognitive scientists believe analogy making is the the core to reaching human level intelligence.
- it develops its own internal simulation engine similar to a human mind.
- it can do dozens of different tasks instead of just one thing, and not just games. State of the art in 2021 is Deepmind trained a computer to play chess, Atari,chess, shogi on a single model, but most computer models are still trained to do a single task.