With a Computer Science Degree, an Old Man At 35?
I think the main thing is that you should love software and technology, not just as a career but as a way of life. That doesn't mean you can't have other interests. But you have to be interested in programming as more than just a way to make a living. Unfortunately this often translates as "no life, willing to work long hours for no extra pay". But not always.
"St Lawrence of Google"
You can build a system that will "make" value judgements based on an analysis of data through the filter of certain programmed assumptions. With a complex enough algorithm and enough input of the right kinds, you can get a sense of world opinion on something. With a robust and mature network of trust in the mix, you can be reasonably certain that the information you get has been tested and should be of use. Especially if it is able to take into account complex contextual input to adjust the information to fit your situation.
In that sense, with those and other tools in place, such a system can advance human knowledge the same way humans do: by analysing and synthesizing existing knowledge. Even Einstein stood on the shoulders of giants. In a sense his intuitive guesses were based on disregarding prevalent assumptions while still taking into account existing information.
The main problem I can see with the idea of a system that can just create a new scientific theory is that thorough science involves experimental verification, refinement and reinterpretation of theory. This could be automated and sped up too, I suppose, but it would involve a lot more resources to build than a system that can just make new non-scientific knowledge. And either system would be a lot more difficult to build than one that just provides opinion, nuance and interpretation as useful meta-data for existing knowledge.