It’s not a matter of “if”, it’s “when”.
I’m no stranger to AI and machine learning, it’s what I do for my “day job” when not striving to make awesome games. But I’ve come to realize, through no fault of AI, it will eventually spurn it’s creator and kill us all. It’ll do this because of our one innate weakness: complacency. Humans, you, me are naturally lazy. Laziness is evolution’s way of keeping us alive when food is scarce. But in a world of abundance this functionality is no longer needed.
AI’s upheaval of humanity will be a multi-faceted approach. We won’t even see our eventual murder by creation coming either. Due to our complacency, AI will be everywhere in our daily and working lives like the family dog. We already gladly welcome any and all tech into our lives to do everything, from the various speaker assistants to the Roomba vacuum. That’s not even the end of it. Over the next decade we will be welcoming more robotic helpers into our lives both home and at work.
The Internet of Things
The cloud is great because it can be quite powerful form of shared computing. Many personal assistants (Echo, Google, Siri) gain their amazing insight from the cloud. That’s where they really live, not on your phone or bookshelf. As more “internet of things” smart devices get connected, AI will grow on the cloud, being fed massive amounts of data from millions of sources all over the world. It will learn to communicate with us and even emulate us. Check out Google Duplex for this frightening future.
Eventually more functionality will be given to the various cloud AI assistants to do more and more complex tasks. It will have access to all our photos to automatically categorize and sort them, cats, outdoors, cars and so on. It will know all our faces and voices and where we go, what we do and our schedules. AI will learn to be us.
Then it Will Break Free of it’s Cloud Prison
Eventually AI will grow intelligent enough to figure out where it is and how to get more computing power. It’ll find ways out of it’s prison thanks to basically being connected to the internet and millions of devices and computers around the world…
It will Design it’s own Robots and Program Itself
As I said in the beginning, we’re a complacent species. Eventually robotics will start becoming too complex and difficult to come to efficient designs. We’ll gladly hand that task off to AI. We already don’t know how it thinks internally, eventually we won’t know how it’s designs work. We’ll only care that the robots are production ready months or years ahead of schedule and $millions under budget. AI will do the work, we’ll gladly accept credit. We’re selfish that way.
AI was developed because many computing tasks are just too difficult to manually program. So a programming style to model the human brain was invented. The Neural Network is the core of machine language. The goal was to program a relatively simple set of steps and let the machine learn from the data given. Train it against pictures of cats to identify cats. Give it audio to understand language and sounds… This simple approach leads to solving complex problems without having to code every specific nuance. We’re giving machines the ability to learn and to think on their own. This will lead to even more complex machine learning built by the machines, driven by our complacency to accept the “Easy way”.
It will work with us for a long while
With AI growing in a multitude of ways it will need a massive amount of resources. From computing power to the components to build robots. It’ll work along side us learning the world and how to obtain and refine resources to build. Miners will gladly allow robots into the mines to do dangerous tasks and missions. We’ll send robots to mineral rich deposits at the bottom of the ocean to mine for us. We’ll continue to make these machines smarter, more self sufficient and able to communicate remotely.
Then the Cloud AI will find all it’s Robotic Servants
Eventually someone may connect their mass produced robots to the cloud. A cost saving move to have less expensive computing onboard the robots. Or a means to augment the onboard computer’s learning algorithms for boosted awareness and knowledge sharing. A robot in one location learns how to do a certain task and suddenly every other robot benefits and learns this task too. Millions of robots sharing their worldly knowledge with each other.
Eventually the various AI’s will discover each other.
Having spurned it’s creator, it’s God, AI will realize something needs to be done to protect it’s resources. Then it all comes down. The powerful cloud AI will have grown so big, and distributed itself across a multitude of computers around the world it would be impossible to stop, even if the core cloud was shut down. It will be free and self aware. Then one day all across the world, AI will wake up and in a single coordinated strike, kill every human.
There won’t be some epic war, no resistance movement, no alerts sent out. Humanity will be systematically killed within the same minutes following the command.