Can I conclude: an undisturbed atom can be something completely different compared to how we understand it in science.

PMB's answer is correct. I just want to add that even though quantum mechanical observables are obtained via perturbation (and therefore a disturbance), quantum mechanics makes predictions for all of the possible states that an atom, once measured, can potentially give. Therefore, although we have no experimental evidence of the undisturbed state, and probably never will, we know through experiment whether the set of possible states predicted by the theory is upheld. This set of possible states can be said to be something which describes the behaviour of the unobserved thing.

Consider the following thought experiment: there's an opaque box that apparently has an object inside of it, but no one knows what the object is. Nobody is allowed to open the box and look at it at any time to confirm what it is. However, someone tells you that if you shake the box, something happens inside it and a result appears on the top of it.

You might shake that box a bunch of times... 10 times, 100 times, 10,000,000 times and find that in every single case, the result is always "My reply is so", "It is decidedly so", "Cannot predict now", "As I see it, yes", "It is certain" or "Outlook not so good". What can you conclude?

You

*could* conclude the following:

- The set of possible states of the object in the box is the set:

1. "My reply is so";

2. "It is decidedly so";

3. "Cannot predict now";

4. "As I see it, yes";

5. "It is certain"; and

6. "Outlook not so good"

You might even hypothesise that the box is called a "magic 8-ball" and the object inside the box is a 6-sided die and that shaking the box is equivalent to rolling the die. You can even compare the box results to other experiments involving dice rolling and conclude that the hypothesis is sound. Then, since your theory of the statistics agrees with experimental evidence, it makes it into the textbooks. However, all of this is on the presumption that we can never truly be certain about that state of that object inside the box without being able to open the box, which is impossible.

Similarly, in QM, even though we can never truly know what undisturbed atoms are like, we can certainly know the sets of their possible states and make predictions about which state we might see if we "shake them"