I had a discussion like this with a seasoned professor. I think the best and also simplest thing to do when tackling a new problem is to first figure out multiple ways you think you could solve it WITHOUT doing a literature review. That shouldn't take too long, and doesn't involve actually doing any research, just some thinking and maybe some basic reading (but not a literature review). Then, review the literature, at which point you'll see if your ideas have been tried before and \ or have merit, based on your improved understanding.
What happens a lot in science is that people acquire a lot of knowledge in a field, and then often creativity is stifled, as every solution to a problem is seen in the light of the existing knowledge. It's hard to think originally about a problem when you've seen lots of other attempts to solve it. I call this the 'burden of knowledge'. So I would encourage contemplating a problem yourself before seeing what other people have done. This may lead to an original line of inquiry. But preferably before you spend a lot of time doing real research, do a thorough literature review. You'll also find that contemplating how to solve the problem yourself will help you understand the efforts of others better.