The victory of Deep Blue against Kasparov in 1997 and the victory of Alpha Dog over Lee Sedol in 2016 set off the trend of artificial intelligence, however, artificial intelligence still has so far a huge gap behind humans in terms of intelligence level and range of capabilities, which hinders its development. The reason is that there is an unbridgeable boundary between human intelligence and machine intelligence. In this context, machines that rely on symbols pointing to objects can only perform formal calculations in a closed environment but cannot realize intentional calculations in an open environment like humans. Therefore, the complementary advantages of man and machine are particularly important. The current relationship between man and machine is mainly a hard function allocation, whereas the relationship between man and machine in the future digital world would more likely be a flexible division of capabilities. This paper discusses in detail the bottlenecks of artificial intelligence, the first principle of intelligence, and other issues, and gives more in-depth thinking about human-machine problem amid these problems. It is proposed that human-machine hybrid intelligence is the future development direction of artificial intelligence.

artificial intelligence  /  function distribution  /  suanji  /  human-machine hybrid intelligence