One of the long standing problems of physics is the measurement problem. This is typically framed as the problem that when a quantum system is unobserved it evolves according to the Schrodinger equation which allows for it to be in multiple states at once and yet when we observe any system it is always in one and only one state. Others have noted that the concept of "observation" is vague.
We can refine the notion of observation. Assuming that observation involves seeing, we may ask when seeing occurs. It does not occur when photons hit the retina, or when they are converted into electrochemical pulses. Seeing occurs when electrochemical pulses arrive in the brain and are processed into a visual experience. And this takes an appreciable time to occur after the photon has hit the retina. Moreover seeing is allostatic, the brain is constantly projecting what it expects to see and what we actually see is partially determined by this expectation. Comprehending what we see is also distinct from the event of visual representation.
So the act of seeing is passive in the sense that one must wait for a photon to hit the retina and active in that vision is a representation within the brain and affected by the brain's predictions. In no case does the act of seeing involve physically interacting with a system of interest. If we merely look at a system we cannot change it or actively extract information from it. In order to see an object or system we must wait for photons to be emitted or bounced off it and end up in our eyes. Or we must do something indirect that subsequently results in photons entering our eyes.
By this logic merely looking at a quantum system, for example, cannot do anything as extravagant as "collapsing the wave function". Indeed the causality must be the other way around, i.e. collapse of the wave function must precede the observation. Whatever it may be that collapses the wave function, if that is what is physically happening, this must happen before the system can be observed. Information comes out of the system only after the collapse of the wave function, in the form of waves or particles (depending on how the experiment is set up) that are produced by the collapse of the wave function.
In terms of quantum systems, of course, we cannot directly observe them in any case. Rather we have to do something like forcing a physical interaction with the system such as exposing it to electromagnetic radiation or particles. We then indirectly measure some physical property of the radiation or particles that emerge from the system and then turn that into something observable - a long chain of causal interactions that results in photons hitting our retina and eventual visual representation. Actual seeing and visual cognition is always an appreciable amount of time behind the physical interaction. And thus cannot be involved in the collapse of the wave function, because it happens after the fact.
If it makes sense to talk in terms of collapsing wave functions, there can be no question that such a collapse is caused by observation in any human sense. Rather it is caused by physical interactions.
We often state the measurement problem as though some particle or system is in a state of non-interaction with the universe and is then brought into interaction. But this is nonsensical. In our universe there is constant interaction of quantum fields to keep the universe in existence: a "particle" is always under the influence of gravitational and electromagnetic fields because they have value at every point in space. The isolated particle is a spherical cow, i.e. an approximation that never exists in practice. How then can any system evolve according to the Schrodinger equation as though it is not interacting with the universe? How can we talk about anything in isolation when such isolation is never real?
I'm sure other people have thought of these things before.