Artificial Intelligence: Making It More Human

Artificial Intelligence or AI is the science of making software that reasons about the world around it.

Software that selects music you like is an example. As are self driving cars.

Last month California became the third state in the world to legalise self driving vehicles. Nevada and Florida are the other two.

GM predicts partially autonomous vehicles by 2015, fully autonomous by 2020. The IEEE predicts 75% of vehicles will be fully autonomous by 2040.

Google’s Driverless Car Project has already racked up almost 500,000 kilometers.

Besides being cool, self driving cars may prevent the 30,000 vehicle-related deaths in the US each year. In Australia there were 1,300 such deaths in 2011.

My son turned 16 on Friday. Today he passed his learners license and will shortly be out there on the road. Would I prefer him to be in an autonomous AI-driven vehicle? Absolutely.

Another example of the use of AI is the autonomous vacuum cleaner. Developed a decade ago, the Roomba is a wonderful feat of engineering. But its level of intellect and logic is that of an insect. Move left, suck, move right, suck. It lacks higher order human understanding.

Today, using the same type of AI, we may be able to engineer AI robots and intelligent agents to serve the growing population of elderly people, but imagine being attended to by an insect. All people require companionship and that would be relegating our elders to a soulless existence.

Let me use a traffic analogy to explain my reasoning. When I first drove north up the M6 motorway from London, I was struck by how soulless an experience it was. Sure the highway is a feat of engineering, but it is essentially a slab of concrete with a horde of trucks hurtling along. By contrast, the meandering country lanes of middle England seem to retain the narrative of the country.

And this is the key difference between insect, or animal logic and the higher order logic of humans: narrative and storytelling.

We are the only species that uses an inner symbolic language, which gives us unique storytelling capabilities.

Stories, after all, are what drive human education – think of your childhood fairy tales, legal and historical case studies and even lessons of success in business.

We also uniquely have a strong sense of directed perception. Our inner language can direct our vision systems to answer common sense questions about real and imagined events.

Take an example used by MIT Professor Patrick Winston – he is assembling a table or band saw with a friend. The friend suggests he never wear gloves when operating the saw. In Patrick’s mind he constructs an image of what would happen if a glove got caught in the rapidly revolving blad of the saw. Not pretty!

That image, as story, is enough to deter him from ever operating the saw with a glove. He did not need to read a manual or be instructed in using the saw to figure that out. Nor did he need to see an actual image of a hand caught in a band saw.

Intelligent agents like Siri are fantastic, but we have only started the journey to get to the point where these agents have enough human-like thinking capabilities to be a true companion.