Let me first make a distinction between the different kinds of robots. The most intuitive ones are robots from the Isaac Asimov school of thought—human-shaped robots built for strength, meant to help humans scale their lives better. These are appealing at first thought. Who wouldn’t want a humanlike companions who could take care of the house while we go about working? Yet all evidence shows that these won’t be the kind of robots that will first permeate the world. They’re too ambitious. It’s like wishing your 5-year-old son would become Roger Federer.
The other kinds of robots are swarm-bots. These are more like ants and bees, working collaboratively. Replicating animals seems a degree simpler than replicating humans. Yet, these are again too ambitious to begin with. Of course, studies and prototypes of these robots are a good exercise in building a knowledge base.
Then there are the simpler kinds of robots that are closer to the artificial intelligence spectrum than the full-grown robot spectrum. For example, there are self-learning thermostats, 3D-visualizing smartphones, self-driving cars and the like. They are more like machines embedded with intelligence than self-sufficient robots.
A curious way to understand these is to look at them as smaller functional units of the full-grown robot. The reason these shall permeate the human world faster is the age-old phenomena of evolution and growth. In this case, it isn’t the growth of the machine, but the growth of the human intelligence to produce such robots. We must learn by breaking the complex Asimovian robot into smaller parts. First we must create and inculcate the smaller functional units and then go on to the more complex assimilations of these units.
Last week the Computer Science department hosted a talk by Erik Demaine, professor of computer science at Massachusetts Institute of Technology. Demaine is a child prodigy who completed his Ph. D. at 20 years old and became the youngest professor at MIT. He also solves cool problems.
His work is centered on exploring and mastering only small functions of robots. He works on finding optimal solutions to getting one arrangement of blocks to take the shape of another (a function of the more complex assimilation we call “transformers”). He tries to find optimal ways to make the bots on the boundary of a swarm of bots understand how to expand the boundaries. He solves the smaller functions one by one, hoping he will be joined in time to recreate his inspirations that range from the swarm-bots in The Matrix to replicators shown in Star Trek. There is also other evidence that suggests this gradual progression.
The smaller functional units are the need of the hour. Sure, robots are cool, but there is no evident reason for the economy to want metallic figures running around streets, carrying your groceries. But small thermostats that intelligently manipulate the temperature at your house and keep your electricity bills in check, sure, here’s my money.
Google released another visionary project recently. Tango is a small device that it built in collaboration with a horde of artificial intelligence companies such as Movidius. It’s a smartphone-like device that lets users build applications that use its advanced infrared imaging and camera systems. A single android app can then be built to scan a room and build a 3-D projection of it, complete with measurements. A small functional unit of the Asimovian robot will be adopted in the immediate market.
The Internet of Things is another ingredient contributing to this distributed revolution. Giving intelligence to smaller pieces of hardware is generally not enough. Getting these small devices to communicate is huge. Even as IPv6 facilitates the Internet of Things, it also enables these smaller functional units to collaborate—a feature that will be key to having more complex assimilations.
The hardware renaissance in waiting is making the development of these simpler, artificially intelligent machines gain more potential. Lower prices of silicon, due to cheap Asian manufacturing, are acting as a catalyst. If only hardware found a more efficient way of distribution, like software, it would gain the same permeability as the latter.
Slowly but surely, Asimov shall have his chance to say “I told you so!”
