Recently I read news articles about a major tech company investing in robotics. Normally I would be excited about the details and dive into the article, but this time, I was not. Instead, disappointment. I caught myself saying “Wrong, wrong wrong! Such a waste.” This is such an unusual position for me to take, and thus I decided to explore these feelings in writing.
It’s not that they were investing in robots, it’s that it was all the wrong kinds. I saw several companies all specializing in advanced robots that mimic people. Walking robots, talking robots, the works. How could this possibly be a bad thing?
Before I address the question, I would like to bring the reader to another class of extremely successful robots. Just about every business, large or small, owns one of these. In a large part of the world, it’s safe to assume there’s one of these in nearly every home. These robots excel at one task… they take stacks of paper, draw arbitrary computer fed patterns on the paper, and present the paper to the owner. Some are even multifunctional robots, who can also fax, scan, email, or copy documents as well. What they all have in common is that they have absolutely nothing to do with people’s standard perception of what a robot is. Neither people nor printers would benefit from printers having human-like characteristics. This is so much the case that people generally don’t consider printers to be robots at all. However, the truth is that robots have been happily working in offices and homes for at least a generation now and everyone has forgotten this obvious fact.
This never used to be the case. Some early printers mimicked humans punching on manual typewriters, complete with typewriter ribbon and mechanical strikers. Other printers, called pen plotters, mimicked humans holding pens. These automatic pen machines would draw in the same fashion a person would, just faster, and with better accuracy. Printers today don’t resemble either typewriters or pen plotters, but rather, they exist as boxes with very complicated internals, that spit out documents that we need. The mechanisms, usually either ink jet print heads or laser drums, have come a long way from automated typewriters or pen machines. The truth is that while people may be pretty good at doing a lot of things passably, any kind of efficiency or specialization of machinery usually ends up looking nothing like a human being.
Add to this, the incredible complexity and expense of making a human-like machine, and these kinds of contraptions become even less appealing when a small box a thousandth the cost can do a far better job. It doesn’t make sense. Machines shouldn’t be any more human than they need to be.
The argument from the other side is that human-like machines can be flexible in their tasks, and often times that ability to do just about anything is paramount. The idea of the human-like machine doing work the same way as a factory worker or butler or farmer does has been around since the very first dream of robots. Obviously, general purpose machines would have a clear benefit to mankind by removing the toil of manual labour. They would change everything. Looking at the problem this way, it’s easy to understand the investment in their creation.
Rather than applaud research in this direction, I see it as lacking imagination. Why would a general purpose machine that can do anything we ask of it need to look like a person? Even modified with extra arms or head or whatever… why? In my opinion, following the principle of just about every commercially successful robot ever made, human-like robots are completely irrelevant. What is needed is a complete re-design from scratch, to define from basics what it means to do general purpose work, and start from there. To do that, following the principle of not dragging our humanity into this, we should throw out our human notions of what general-purpose work should look like in the first place.
How do we define general purpose work? For most people the obvious answer is a good pair of hands working hard and the mobility to move around. And that description is obviously wrong. Hands are handy, but totally useless compared to even the most basic of tools. How far would a carpenter get without his tools? Most people wouldn’t even get through dinner without putting at least one tool in their hand. Even stone-age people knew the limitations of their hands as they picked up sticks and rocks and made tools out of them. Hands aren’t so handy as much more than the general purpose connectors to the tools that we use. So the true obvious answer isn’t hands as much as it is the handy access to a variety of tools suited for job at hand. That’s the real definition.
Now how about the other part of the equation… mobility. Even compared to the animal kingdom, people aren’t particularly mobile in a useful way. Most animals that need to be mobile do it better than we do. Similarly in the stone age, people built rafts, made shoes, and rolled heavy objects on logs because they understood how limited their bodies were compared to even the simplest machines made from stone, wood, and animal parts. Where would civilization be without the wheel? Rather than spend exorbitant resources designing and building a machine that walks like a man, slapping a pair of wheels and a motor together and calling it a day can be done a million times over for the same effort. There is no reason to re-invent the wheel when the wheel works fine.
There is a third part to this definition, and that is the know-how needed to do the work. A craftsman isn’t paid for the quality of his tools, he’s paid for how well he uses them. How could a dumb machine compare to the high-tech intelligence of a machine built to look like and work like a man? This question is easier to answer, because robots can’t think. Computer science hasn’t even settled on what the idea of cognition means in the most basic sense, let alone design a machine that can think for itself. Based on this, the idea of designing bodies for machines that we don’t even have a clue how to properly control is foolish and absurd.
So… based on hands and feet not being so useful, and artificial intelligence not even existing yet except in the minds of science fiction writers and gullible investors, spending huge sums of money building stereotypical “robots” is a complete waste. They are literally spending money building machines based on a science that hasn’t been invented yet. They are trying to copy hands and feet that (on their own) weren’t even good enough for cave men. Even chimps have the sense to wave sticks around. Venture capitalists must be a lower form of evolution. This disconnect between fantasy and reality is what drove me to react so negatively to the news reports of companies investing in human-like robots. It’s madness!
So the idea of building general purpose machines is a fool’s errand. Well, at least it is to people who cling to traditional ideas of what robots should be.
Discarding those ideas leads to much more optimistic prospects. A variety of tools at hand is easy to understand to anyone who’s ever changed bits on a screw driver. Mobility is easily solved for anyone who understands wheels and motors. Almost all basic problems have been solved long ago by mechanical and electronic means. Even intelligence is a problem that can be solved only with what exists today. Robots like printers are already about as intelligent as they need to be, they can even respond to issues like low ink or lack of paper in a meaningful way. What computers do well is simple tasks done repetitively. Building a moderate level of intelligence into individual tools is the key. People controlling the system can then automate processes, leading to a startling level of complexity without any large machine intelligence needed. What is needed, is the sense to abandon useless old ideas, and the imagination to invent something new that only relies on what is readily available. We don’t need science fiction ideas, we need to put the tools we have to proper use. Rather than bring to life the robot dreams of the old science fiction masters like Isaac Asimov, we need to invent what we can, and thus bring into the world an amazing reality that science fiction has never even imagined.