We all know that greedy gets us nowhere. We were all probably told by our moms to be kind (e.g., friendly, generous, and helpful), who were probably told by their moms, and so on. This crazy idea may be traced all the way back to some named or unnamed philosophers, but how can it make sense? Why should we hand out our precious resources (e.g., time, things, or even opportunities) to others in this hyper-competitive world? Why shouldn’t we calculate the costs and benefits of all our potential options and pick a move that maximize some sort of utility functions (e.g., money or advancement)? This is precisely what we do when playing chess or tennis, when there is no friends and only one opponent in the game. Nobody expects us to be generous there.
The reason is probably that we are not capable of making many meaningful calculations in life. Beyond a few artificially constrained games, we are severely under-actuated and underpowered creatures that are trying to navigate in the vast ocean of human society. Each decision that we made may not change that much how we move up or down in a long term. Instead, the movement of waves below us makes far greater a difference. With a very limited horizon, we have no way to know for sure what’s around us and what’s coming up next. How can we make a decision then? Moms told us to use heuristics that have been proven to make long-term stochastic sense (ok, not in these exact phrases), which are to be kind, friendly, generous, and helpful, among others.
Now, let’s take a minute to think about robots. Our robots today are greedy. They are self-interested, having a tunnel vision (not literately) of the world around them, and trying to maximize some sort of utility functions. They work well in structured environments that can be fully modeled, and are getting better by days in more complex settings. If we use a linear interpolation to predict the future, the robots will get smarter, more capable, and more selfish. This is probably why the internet is full of worries about our future with robots.
I think this is like saying all chess and tennis players are greedy, and we should be careful with them. The robots today are self-centered only because the way their working environment are set up to be. As moms of robots, we, roboticists have to teach our robots how to survive in the real complex world, a world that greedy gets them nowhere.
If the robots are going to be as intelligent as humans in the future, we should not expect them to be that different from us: there will be good robots, there will be bad ones, and there will be many in between ones. And all these robots have to deal with good, bad, and in-between humans. It’s this enormous diversity that invalidates short-term and self-centric thinking, and makes it more important to be kind to others. Call me wishful thinking, and I don’t know if we should feel happy or sad about this, but the kind robots are probably the ones that eventually will replace us.