Complexity

Artificial Intelligence is hailed as a tool and method of infinite potential. But AI in its current form does not tell us much about ourselves as humans - but science does. Current AI is engineering, it is not human discovery. AI cannot teach us about ourselves more than we know. AI is about doing whereas science is about knowing. AI is predictive while science is revealing. Science answers the why, while AI lingers on the what. AI automates existing processes in a fast way. A human (or group thereof) can always outperform AI given enough time. AI displays/copies human bias learned from human data - it doesn’t even rise above the collective view. Human knowledge grows by asking questions, AI doesn’t ask questions.

Some critics of evolution, claim there was not enough time for the complex human gene to evolve via natural selection. The human genome contains 3bn base pairs and 30,000 genes. If we think of DNA as a word of length L and the bases (ACTG) as the different possible letters of size K (4), then we would have L^K possibilities. This number grows very quickly.  One mistake of using L^K is that we assume one character changes at a time, but this doesn't need to be so. Mulitple characters can change with each offspring or generation. We can do a little computational test. Let's say we want to guess a 6 letter word out of 5 different letters (abcde). This would give use 15,625 different possibilities (5^6). A random search roughly takes 15k steps to find the right solution. But what if once we have learned a correct letter we don't change it anymore. How much does the search reduce by? Below you can see the code for this, we have a function gen_seq which randomly guesses string...