A Look at the Future
Already, autonomous implementations of AI are common. If you are on Facebook, for example, software decides which posts are going to appear on your feed. Financial traders execute buy and sell orders for hundreds of millions of dollars only nanoseconds after a market change has signalled that it is profitable to do so. When you buy a can of beans at the supermarket, scanning it at the automated checkout may trigger a buy order to a supplier. In turn that may generate instructions to run a production batch, move the stock from the warehouse to despatch and schedule a delivery to the supermarket, all without human intervention.
But that’s only the beginning. The examples I gave show AI executing instructions using systems that humans designed, with products that humans invented and designed, and doing work that humans would have.
The next step is AI systems that develop themselves, faster and better than human programmers could.
Remember that AI is becoming exponentially smarter and smarter. Cybernetics analysts claim that AI power doubles every 18 months. Its repertoire of abilities increases and so does its efficiency. In twenty or thirty years it will be ten thousand times more powerful than it is now, smarter than any human, and may even acquire the phenomenon we call consciousness.
In the not too distant future, AI won’t need humans, except of course as the end consumers of some of the products and services that it creates. Autons will start inventing their own new products; moments later the product will be fully designed and ready for consumer testing. And if the consumer is another Auton, acceptance and the confirmation of a multi-million dollar deal could be only another second down the line.
Autons will design the machines that make the products, build the production plants, and manage them. Autons won’t even have human owners.
The big obstacle in the way of this happening is ethics.