Consciousness forges our perception of the universe, not the universe itself.
"the quantum physicist's radical new take on reality seems to suggest that the world we see emerges from our observations." This is not new. It was obvious to the ancient Greek philosophers. The problem is, physicists have failed to grasp the distinction that the "world we see" is not the same sort of thing as the "world as it is."
"Our theory says that only observations are fundamentally 'real'." I think, therefore I am: Our thoughts are the only thing that we ever perceive. But what we think nature is, and what it actually is, are two very different things.
"questioning one of our most basic assumptions that we operate on as human beings: that reality is objective and pre-dates us." Our reality is entirely subjective - a virtual reality created entirely by our brain - that, among other things, attempts to construct models of reality, that physicists subsequently confuse for reality itself.
"For example, before we measure a quantum system, it can hold contradictory properties" No. It merely seems contradictory, because the constructed model fails to correctly account for the behavior of entities that manifest only a single-bit-of-information.
"When we observe it, we force the system to assume a particular state" No. We only force our model of it, not the thing itself, into a particular state. A coin does not suddenly become one-sided, simply because an observer "calls it" either "heads" or "tails".
"Can that give you something that allows you to predict what you'll see" No. Deductive logic can only derive conclusions that follow from some premises. But it cannot establish the validity of the premises. Information, in Shannon's Information Theory, is, in essence, defined to be "that which is not predictable."
"algorithmic information theory will find many applications in physics in the future" True. But it is Shannon's original Information Theory, that really explains the nature of quantum theory, not the later-developed, algorithmic information theory.
"Surely the defining feature of agents should be that they act?" And the most fundamental possible action, is to simply detect the existence of something else- by interacting with it. That is what single-bit detection, is all about.
"Choosing one's actions is a secondary notion that should not be a fundamental part of any physical theory" No. You do not get to choose to act with something, until AFTER that something has been detected. "choosing" what to detect in the first-place, DEFINES the things you can interact with.
"radically different approaches are, in my opinion, desperately needed." I agree. But this one is not radical enough.