Algorithms are not magic - Humans define them as sets of rules which can be simple or very complex (obscured). They can be used for good or bad (like most things). Machine learning is still based on algorithms but the computer is able to very quickly test the existing algorithms (assumptions) and compare them and make guesses at what are likely new algorithms. The computer will either test and get validation (success and save that new algorithm) or a human may tell it to discard that algorithm as bad/failure/etc (like a parent tells a child).
Computers don't yet have any real intelligence or feelings - they can just compare vast amounts of data and algorithms very quickly without mistakes. Humans also compare what they know mentally (which is sometimes incorrectly based on bias, incomplete comparisons, or opinion) and will come up with a new theory or guess. Humans do have an intuition as an advantage but to be honest it's not always correct as a scientist may be more accurate than others around science, whilst a musician or artist may be better for other things. If we feed computers with incorrect algorithms we can be sure they'll give biased or incorrect results. We teach them, and they compute for us.
Fear the programmer, not the Computer!
See What is an algorithm, anyway?
What you need to know about the simple concept that powers the modern world.