There are infinite classes of conjectures that cannot be disproven, and as such they are not useful in the sense of adding anything predictive about our understanding of reality.
Science is about asking questions of reality, via experiment, as to what its nature is.
And we need to be clear about how science deals with proof. Science does not get absolute answers.
Science is about dealing with uncertainty, eternally, and reducing that uncertainty where possible.
Science is much like the Japanese concept of kaizen – continual improvement, without ever reaching perfection.
In science, we achieve that by looking at the available explanatory frameworks, and seeing where their predictions are sufficiently different in certain contexts that we can distinguish between them, then we design experiments to test which of the available explanatory frameworks best fits the results.
And we have to accept that mathematics contains classes of systems that are fundamentally unpredictable for many different sorts of reasons, and reality seems to contain instances of those classes of systems. So we have to accept that some things may not be predictable, even in theory, let alone in practice.
This applies right down to the fundamental axioms of understanding.
Some people believe that all things are related by necessary and certain causes.
Others of us can accept the possibility that fundamental uncertainty, if it is constrained within probabilistic bounds, can in aggregate approximate causal systems very closely.
Many people have a great deal of difficulty accepting that latter possibility, even when shown mathematical and logical proofs that such systems are possible.
So there are many interpretations of what science is, and what it does; and being able to falsify things via well designed experiment is one of the major paths by which science extends the ability to offer useful understandings of new levels of systems and structure.
And we have vast sets of evidence already that clearly demonstrate that reality is far more complex than any computational system (human mind or Artificial General Intelligence) can deal with in detail; so any intelligence is going to have to resort to using contextually useful shortcuts (heuristics) in practice, to make any sort of sense of anything complex in any usefully short time.
So we are all in this mess together, and science, if used with a modicum of humility, seems to be a great tool in helping us gain sufficient understanding to find cooperative ways to survive with as much freedom and diversity as possible.