We did not 'create' the curvature of light around the sun that was the first experiment to validate General Relativity. The theory predicted that if we look at the right time, we will see it happen. Just looking was the experiment, with no need to create a specific condition. And the shift itself was measured, even though it is thoroughly beyond our control.
Social and psychological experiments are sometimes just as hard to stage, but we can get data from enough people, with enough clarity, to statistically validate a hypothesis. Answering the question "Are men or women more likely to be criminals?" from historical incarceration data and self-report of crime victims, is still an experiment.
So this characterization of an experiment is a bit off-center. As someone noted above, if it were this absolute, we would not have a science of meteorology, we would just have almanacs.
Thus the first question is surely a 'no'.
The reverse question seems equally simple. A Geiger counter is an example.
Radioactive material throws off alpha particles at an expected average rate, but in a truly random and utterly unpredictable rhythm, due, at root, to the absolute indeterminacy of the real distances between nuclear particles. The time between counter clicks will vary in a truly random way. There are even important laws like Heisenberg's principle that only ever give us ways to make things unknown and unknowable at some level of detail.