The theory of evolution seems to be intelligent as it creates life
When you say "seems to be intelligent" that begs the question: How are you defining "intelligent"? Which of course is still one of the big issues in AI research.
I think there are some flaws with the argument that "creates life" = "intelligent":
Evolution does not create life. It operates on entities where there is a copy mechanism which is not 100% reliable, plus a selective environment that also impacts likelihood of further copies being made. Some form of proto-life (an initial Darwinian ancestor or Ida) needed to exist before evolution started.
The process of creating the first proto-life capable of undergoing evolution is generally thought to be a large semi-random search through chemical combinations. Random search is sometimes used in optimisation problems, and might be studied as part of AI search topics. However, it would normally be considered something of a baseline algorithm, and definitley not tick the boxes for all the general traits of intelligence.
Isn't this evolutionary mechanism itself the same as the essence of human intelligence?
The Wikipedia article on artificial intellegence lists challenges faced by researchers and developers in AI. The categories chosen there are:
Reasoning, problem solving; Knowledge representation; Planning; Learning; Natural language processing; Perception; Motion and manipulation; Social intelligence; General intelligence
Together, these are mainly traits of mammalian, avian and a few other multicellular species, with a few traits such as language heavily focused on humans.
I think it is important to separate out the mechanism whereby these traits arose naturally - which is generally agreed to be via an evolutionary process - from how those traits function. Artifical intelligence may use a little bit of reverse engineering from the natural traits in order to inspire design, but most AI systems do not use theory of evolution directly.
When used directly, evolutionary algorithms can be used to solve search and optimisation problems. Also they can be used to solve simplified problems in perception and motion/manipulation. However, we are not able to scale up such algorithms to solve all aspects of general intelligence. Instead, systems like machine learning are designed to work from analysis of the problem, inspired in part by working natural systems. These work far more efficiently than evolutionary algorithms. There are no competitive evolutionary variants of AlphaZero, Watson, GPT-3 or neural-networks used in image processing.
Evolutionary algorithms have their place in AI in practice and research. However, they do not define or encapsulate a form of general intelligence.