Re: "a real scientist is sceptical of ANYTHING that is not proven"
I'll even go even further by stating that several scientific advances happened when a few scientists were sceptical of something that was supposed to be proven.
An ideal scientist, in the modern sense, is sceptical1 of everything - or more precisely, acts as a Bayesian reasoner in regard to every hypothesis. Per Descartes' "evil genius" and the like, an ideal scientist knows that his or her own sensory impressions and/or process of reasoning might be corrupted and lead to incorrect conclusions, and thus no purported fact should ever be treated as having probability 1.
In practice, of course, it's impossible to operate without a great many axiomatic assumptions, and ubiquitous vigilance isn't feasible either. So actual scientists have to relax their sceptism many times a day. But scientific epistemology works best when its practitioners start with the explicit assumption that everything is open to question, then decide which hypotheses to provisionally accept based on their high Bayesian probability. So, for example, a scientist will typically decide to trust sensory impressions at some level (for example when reading the output of scientific instruments); to assume mathematics is consistent; to assume reasoning that seems to be logical in fact is so; and so on.
But "proven"? That's faith, and is fundamentally non-scientific. So is "disproven".2 What an ideal scientist refers to as "proof" is really only one of two things: the description of a tautological result (eg in mathematics3), or a gloss for "looks like a really high probability of being true, so we'll treat it as true and move on to other things".
1Or skeptical, if you prefer. Either spelling is etymologically justifiable.
2Fans of Karl Popper think disproof, or falsification, has special status in scientific epistemology, but that's an inferior model. Under a Bayesian interpretation of scientific epistemology there's no need to give falsification special status. A falsifying result shifts the probability of a hypothesis substantially downward, and that follows from the math, so there's no need to treat it differently from a confirmational result. Any amount of falsification only approaches a probability of 0 for the hypothesis asymptotically, but that's also appropriate, because it's always remotely possible that every single experiment was flawed, for example, so no hypothesis can be rejected absolutely. (See again the evil genius problem.)
3And that's still subject to interpretive error, distortion of consciousness, etc.