If you would have asked the child version of me what the most important job of a robot, I would not have hesitated. Not for a second! All my friends knew the answer too. This seemed so obvious: the most important job of a robot is to sense upcoming peril and flail its arms, yelling, “Danger, danger!” or even tagging someone’s name on the end instead of another “Danger!” Example: “Danger, Rob Williamson!”

Actually, many decades later, we know that most robots can’t really anticipate a thing. They’re not supposed to. The first industrial robots came out somewhere around 1960 and they could do one thing, whatever that thing was (the very first robot stacked iron castings and it was strong but not so intelligent).

Many robots are the same today but what saves their jobs is that they are much more flexible today. Not only are they versatile, it’s easy to keep them that way because the software that controls them is really versatile, and can partially write itself. This really has nothing to do with the robot, though—it has everything to do with the software.

Robots have been around since 1770, when a device called The Mechanical Turk would play chess against opponents and win just about every time. It turned out there was a chess master hiding in the chess desk, moving levers to reposition the chess pieces. A scam! It wasn’t a robot (or as it was billed, an “automaton”) at all.

Real robots were put to work immediately, that’s what they do best. But our relationships with them must start before the work starts. Even the most advanced cobots need us to show them their universe, where they might hit something or someone. Then we need to either train them with the motion we expect, or program their movements. At least the programming task has become much more code-slinger friendly, as noted above.

It is entirely possible that we will have to change the actions of the people that the robot or cobot will be near, work with, or work on something in team fashion, interactively. They will have to be careful in the vicinity of these helpmates, unless of course these robots come with sensors and software that tell it to slow down when a human nears, and stop when a human touches. The point is, the robots will have to act differently and so will we. In the meantime, those robots are doing the dull, dirty and dangerous work that we have been looking to offload for years.

So long as we don’t look to the typical manufacturing robot to anticipate things and change its behavior accordingly, I think we’ll be fine. However, we must admit to, and plan for, an “onboarding” process for our automated friends.