This can be filtered to some extent - indeed modern phone cameras already do so quite heavily, to overcome the limitations of their already small pixel size and already high sensitivity. But doing so costs the same detail you'd need to try to invent a "simulated telephoto" image, which would give your fake telephoto process even less to work with than otherwise - guessing you have a neural net in mind here, and while I'm not about to argue that a sufficiently well-trained net won't produce some kind of result given an input of sufficient similarity to its training set, I see no reason to expect that result to bear any particularly photographic similarity to the original input.

I mean, don't get me wrong - what you seem to be suggesting isn't all that dissimilar from how we currently understand the human brain's own optical system to work. But I don't think it is especially likely that many such brains will happily accept another neural network's best-effort guesses as the output of a process that we've all learned to expect will give us representations as precise as is within the capabilities of the devices we use to make them.