The Insight Source Blog
Your preferred Telematics information resource
Author: Lance Holt
Content Marketing Specialist
Put Your Drivers in Safe Hands with Vehicle GPS Tracking
Improve the Safety of Your Employees and the Public
Foris Solutions Increases Driver Safety and Decreases Fuel Costs
When Machines Take Over
How Far Away Are Autonomous Vehicles?
Autonomous vehicles and artificial intelligence (AI) have been a science fiction dream for a long time. In some ways, they still are specters of our imagination, but how far away is this dreamworld of autonomous vehicles for consumers and commercial vehicles?
For consumers, this fantasy isn’t as far off as one might think. Will it happen in the next five to ten years? Most likely not. There is a laundry list of reasons from a safety aspect alone that people working on this project still have to sort out, but will it take 100 years? Again, most likely not.
Even though we might be closer than you think to see autonomous vehicles on the roads for consumers, we are still a long way away from companies implementing them into their mobile workforce and being any kind of a threat to GPS tracking companies. Let’s take a look at recent advancements in AI and what some of the roadblocks are that need to be overcome for autonomous vehicles to become a reality.
Advancements in Artificial Intelligence
The largest piece to this puzzle is the AI aspect of autonomous vehicles. Unless you pay close attention to this field, you may not realize the advances that are being made. AlphaZero, an AI program developed by Google’s sister company, DeepMind, is the best illustration of this. AlphaZero was given basic rules to chess, like how certain pieces move, but not a strategy. Using what’s known in the AI world as reinforcement learning – playing against itself over and over again – AlphaZero taught itself chess in four hours and went on to defeat the best chess player in the world. The chess player that AlphaZero bested was another AI program, Stockfish 8.
We also see evidence of the significant advances in AI with Tesla’s autopilot system, or as some have called, semi-autonomous vehicles. The first fatality that happened while autopilot mode was engaged did not occur until after 130 million miles were driven on autopilot across all Tesla vehicles. The important factor in this incident was that the accident was not due to the autopilot mode malfunctioning, but rather, the failure of the person to react to the situation. When you consider this fact in comparison to the nearly 1.3 million car crash fatalities that happen on the road every year, Tesla’s semi-autonomous vehicles has been incredibly safe and reliable.
While Tesla’s semi-autonomous vehicles are proving their fundamental safety, there are still philosophical problems to address before handing the reigns completely over to the computer in your vehicle. One problem to think about is how to input human common-sense into a computer. This is illustrated by considering what happens when you get into a self-driving car and tell the computer, “get me to the airport as fast as possible!” and then 10 minutes later you arrive at your departure gate completely disheveled. The AI didn’t understand the human common-sense factor that when you say something like, “get me there as fast as you can,” you still want it (the AI) to consider your safety and well-being, not to mention you don’t want it driving so fast that it forces you to lose your lunch in the process.
Another and more disturbing problem to consider is what happens when an autonomous vehicle is driving in a residential area and a child runs out into the street. Whose safety does the AI covet over all others? Does the car swerve into a mailbox saving the child, but potentially severely or fatally injuring the passenger in the vehicle? Does it swerve onto the sidewalk avoiding the child, missing the mailbox to protect the passenger, but potentially harming the five bystanders walking past at that moment? These are safety concerns we must think about and decide how we want to address in self-driving cars before they become commonplace in the consumer and commercial markets.
Are We There Yet?
So, how far away are self-driving cars? After considering what must be done to make this long-standing science fiction dream a reality, it’s not as far off as one might think for consumers. We already have semi-autonomous vehicles on the road in Tesla’s autopilot mode. We are even further away from autonomous vehicles being a threat to GPS tracking companies. Autonomous vehicles will have to be adopted by consumers first before companies start implementing them into their mobile workforce.
Before we worry about Skynet and killer machines taking over, take solace in a quote from Garry Kasparov, a chess grandmaster considered to be the greatest of all time. Mr. Kasparov was the first human to lose to an AI computer developed by IBM called Deep Blue in a six-game match in 1997. While on the podcast “Waking Up” by Sam Harris, Mr. Kasparov stated, “Worrying about killer AI is like worrying about overcrowding on Mars.”