The exponential growth of AI

The exponential growth of AI

AI brain with numbers and a chip

Comments

The Exponential growth of AI

Over past decades, the computing power behind machines has doubled every two years, this is known as Moore's Law. The plain truth is that computers are getting better, faster than anything else ever. A Sony PlayStation4 (PS4) or a Microsoft XBox1 are today more powerful than the most expensive military super-computer from 20 years ago. As computing power has doubled every two years, the power of machines to do human tasks also increased at the same rate. This kind of growth is called exponential. Most of us are not very comfortable or intuitive when it comes to what we call exponential growth. Exponential growth is constant doubling, 1, 2, 4 8 16.... It multiplies on each of the previous steps, and that means that at first the effects can be relatively small but then it suddenly takes off, and that inevitably catches us off guard. These changes are happening in a digital world, where it's easier to create and replicate new things than in the physical world.

That's why AI progress is so fast, when we look at some recent examples of the kind of things that computers and AI and robots can do, it really does feel to me like we have just passed a tipping point. We are now just on the other side of it and are in some really new territory. We think this is one of the most significant changes, literally, in human history. We can compare it to the Industrial Revolution when machines started automating physical work. Companies in Britain are already using machines which can teach themselves but they say they won't threaten jobs.

Who is Vicky?

"Good afternoon, you're though to Vicky at Virgin Trains, how can I help?" Virgin Trains uses an artificially intelligent software program to handle customer queries and phone calls. It also reads every customer email and understands them enough to sort them into one of 470 categories. In the future, it will even send them to the right people across the office. It will take a look at what you have said, how you have said it, the order of the words, the words you have used, what comes before and after those words. What makes it clever is that it learns from its mistakes. Staff are teaching it so it doesn't make the same errors again and in that way it learns like a human. The more it sees, the more it learns. The more it associates with what it has seen in the past and then the more accurate it will get. It has cut the time spent opening and sorting emails by 85%. Has your use of AI changed the number of people you have had to have in your team? we asked. "No, that wasn't the aim of bringing in the tool and there are no plans to do that. We are using AI as a tool to free up some time that we would usually use on administrative tasks, to enable us to spend that time on the customer." Could this be the first sign of an artificial intelligence that does our mundane work for us? Are machines that behave exactly like us are still a science fiction dream?

Self Driving Cars

Well we have all heard of the Google driverless car, but there are a lot more, and some like the Mercedes-Benz F 015 are AMAZING!

The biggest advance in AI has been in machines that can cope with the unexpected. The development of self-drive cars utilizes this kind of AI trait and has caused a flurry of development and lots of new cars. As we all know Google already has a fleet of driverless cars in California and the only crashes, they say, have been when humans got it wrong. The UK Government is also investing over £100 million into self-driving vehicles...er cars that drive on the pavement! We will see this one being tested on the streets of Milton Keynes next year. We are at the beginning of a journey to get to a point where we can safely have self-driving cars on the pavements. These pods (not really cars) are designed for short, driverless journeys around Milton Keynes. When you get to the train station, they'll take you home via the pavement at up to 15 miles an hour. Being off the road means the sensors have to be more precise than in a driverless car as we are interacting directly with pedestrians. Pedestrians may stop, they may lie down in front of the pod or even try and jump over it. You don't know what they will do next and they are much more unpredictable than cars. There is still a very long way to go with driverless technology, but it's progressing fast. Do you think we are close now to some kind of tipping point when it comes to self-drive vehicles? Audi, Jaguar and Land Rover are also doing a lot of work. I think there is going to be an explosion in this area very very soon. So if you fast forward ten years, what do you think we are going to see on the roads?

Self Driving Trucks?

Truck drivers do not agree that a machine could do their Jobs. Bob Wain is one of 875,000 people in the UK who drive for a living. He says he wouldn't trust vehicles that could drive themselves. He says "I know that if anything jumps out in front of me, my right foot's going to hit that brake. Even if I have to bounce my head off the windscreen, I'll stop this truck on a sixpence. I don't think a computer's going to have that."
We asked Bob "In America, Google has racked up over a million miles now in their self-drive cars so the technology is there to do it, so do you think that it will happen?"
He replied "For the cars you'll probably get away with it, because they can go anywhere. For a truck it's a completely different ball game too many practicalities will get in the way. How would you know your load's safe? If it did ever get broken into, who's responsible for that? What will happen if it's driverless? You're not going to ring the AI police and say, "Excuse me, I'm a driverless truck, I need your assistance."

Article Written By Numus Software : September 15th, 2015.