Ideas, Randomness, Startup stuff

Most ideas about the future seem far-fetched, unless they happen.

It turns out real Artificial Intelligence, known as strong AI, is tough. We still don't have machines that are conscious and there aren't many signs that it will happen soon. Sorry Skynet. What we do have is improving robotics technology--we are getting better at animating smaller robots and creating them inexpensively. Robots are starting to walk and I own a robot that can mop.

An autonomous self-controlled robot that can fold laundry or pick up my apartment is probably a way off. It turns out, as boring as these tasks are, they are complicated tasks intelligence-wise and that's the main reason they aren't on the near-term horizon. Physically, creating the robot shell is feasible with current technology. It's the brain that's the problem.

There's a twist. The internet has made long distances trivial. Outsourcing a software project to India is trivial. Even more importantly, remotely accessing a device, either a remote PC, or otherwise, is trivial. There are whole areas of the world where there is little opportunity--especially if you're not literate.

An enterprising company could create an inexpensive robot form factor that can be controlled remotely over the internet. Set up offices in the third world and recruit inexpensive labor to drive a fleet of house-cleaning robots over the web. Then sell house-cleaning services at incredibly low cost to First World Countries (via the robot helper) while providing a safe, clean job in the third world and presumably making a nice profit. An interesting consequence is that this creates a financial incentive to gradually build autonomy into the robot. If a single operator can control a greater number of robots by increasing their autonomy, then there will be an incremental path for developing autonomous AI. For example, if a robot can find and pick up a shirt on it's own, the operator could switch back and forth between two robots while each robot finds the next shirt.