One of the reasons it is hard to predict what the future looks like is because much of the future is movable. The thing we are trying to forecast is changed by our attempts to make it real. Many hundreds of years ago, when creative people imagined flying machines, they imagined machines that had wings that flapped. What they imagined did not happen; the deliverable moved to an airplane with fixed wings. In the 1950s we imagined wrist-watch radios on every person in the future, and that did not happen (yet), but instead, the future moved to “radios” in the pocket of every person. Now we feel we don’t want wrist-watch radios. The future moved.

This migration is even more pronounced with some frontiers like “artificial intelligence” or “virtual reality.” When these concepts were first imagined, we had a picture in our mind of what they should be, but as the initial parts of the dream were achieved, they begin to change what we expected at the end. Today our kitchen Alexa will answer spoken questions, play the music we request, and turn the Christmas trees lights on command, but we don’t think it qualifies as AI, yet Alexa would have been called AI 80 years ago by any science fiction writer. AI is in a movable future that constantly shifts as we invent versions of it.

Over time we may end up accepting something as “predicted” that was not in our original dream. Very ambitious dreams that take a generation or two to invent are extremely susceptible to shifting what they are. I suspect we’ll see target shifts in long-expected inventions such as nuclear fusion, genetic cloning, space travel, brain jacks, flying cars, and virtual reality. What we get in these fields is probably not what we imagine right now, but when they do arrive, we’ll feel as if they had been expected.

Part of this gap between what we expect and what we get is due to our lack of imagination. The future is extremely difficult to picture beforehand because technologies are mainly governed by technical constraints we are unaware of initially. And technologies are also shaped by social forces we are forever guessing wrong about. The truth is we don’t know how humans really behave, and the interaction of new tech and old humans continues to surprise us. We tend to use new things in new ways we did not expect. So when we extrapolate to imagine what, say, genetic cloning will be like, we rely on past behaviors, and will often be blinded by old patterns.

But another reason why there is a gap between what we expect and what we get is that we actually change what we want along the way. Technically we can make wrist-watch radios now. In many respects that is what an Apple Watch is. It is a watch you were on your wrist that broadcasts and receives radio signals in multiple bandwidths. But even though the Apple Watch has that capability, most people dont use it as a walkie-talkie radio. We’ve discovered better ways to communicate with one another, which shifts the target of what we want on our wrist.

We think we want a walking humanoid AI robot with a two-eyed face and five-fingered hands, but as we attain bits toward this vision, I am willing to bet that we shift what we are aiming for. I bet Robbie the Robot is not where we land. Science fiction depictions are useful in giving us a target future at the beginning, but we should understand that these targets are movable futures; even when they come true, they are not exactly the future we promised ourselves.