Time to embrace the power of AI
We should prepare for artificial intelligence, not fear it, writes Bill Bennett
OneNet CEO Michael Snowden says people fear that artificial intelligence (AI) will take jobs. But he says while AI will disrupt employment, we face a bigger risk in not embracing the technology.
Artificial intelligence is not new. The idea has been around since computers appeared in the 1950s. Snowden says: “AI is about getting a machine to do what a human can do”. He says although there has been some progress, its history has been one of false starts. Until now: Snowden says the technology has taken off in the past few years.
There is no single reason for AI’s rebirth. Snowden says this technology revolution, like so many others, is due to a confluence of events. “It comes down to a number of processes that have been working in parallel and are now all delivering at the same time”.
First, there’s the long-term effect of Moore’s Law. This is the idea that computer chip performance doubles every two years. It’s been going on since the 1960s.
AI needs a lot of computing power. Previous attempts at AI floundered because computers were not up to the job. Today, computer power is cheap and plentiful enough to handle AI’s massive processing demands.
A decade ago, researchers discovered the graphics processor chips used for video game graphics were ideal for building a specialist AI computer known as a neural network. This can be programmed to learn by example, in much the same way human brains operate. Today Nvidia, the company behind those graphics processors, is one of the hottest names in Silicon Valley as it turns its attention to AI.
Another event was the rise of online search and social media. Snowden says both collect the huge amounts of data needed to fuel machine learning. Companies like Google and Facebook have access to vast amounts of data. They can put that to work teaching machines to understand how our world works.
Meanwhile, new software techniques such as generative adversarial networks have emerged. In effect, these set two AI systems to compete, to improve their learning. It’s a way of teaching machines without huge datasets and gives companies without access to vast pools of data a way to compete.
Better processors, more data and new techniques mean after years of disappointment AI is back with a vengeance. The sector is booming and is set to continue growing at a rapid pace. Overnight it is everywhere. AI is turning up in mobile phones, security software and even in accounting systems.
Snowden says although AI is growing fast and can perform tasks normally left to humans, it won’t necessarily bring mass unemployment.
He blames Hollywood film makers for some of the fear. He says to many people AI is HAL in 2001: A Space Odyssey or the Robocop. “These are examples of artificial general intelligence. We’ll get there one day, but it is still a fair way off”, he says.
Today’s AI is less threatening. Snowden says machine vision is a good example. It is a technology where a machine can look at an image of an animal and decide whether it sees a cat or a dog. The same technology turns up in work done by Google’s Waymo. The autonomous vehicle company uses machine vision to decide if an obstacle is a human or another car.
HThere’s a track record for Snowden’s optimism. He says when cars replaced horses people had the same fears, but jobs emerged building roads, providing fuel and fixing cars. He says: “Coachbuilders turned their skills to making cars. PCs created more jobs than they lost. When banks installed ATMs they kept tellers on to sell other services. Today banks employ more people than they did before ATMs.”
Although the big picture is positive, some groups will suffer. He says many dangerous, repetitive or dirty jobs will go. Professionals won’t be immune. “Think of a lawyer researching cases. They may have to go through hundreds of documents looking for relevant terms. It’s a job that AI can do, but it will more likely augment a lawyer’s work than replace it.” He says market analysts and journalists are also likely to face change.
Yet Snowden says there are things AI can’t do. He says empathy, judgment and creativity are beyond today’s technology and may always be. Leadership is another skill that can’t be programmed.
There will be unexpected effects. “Who would have thought in 2007 when Apple launched the first iPhone ‘there goes the taxi industry’”?
Despite the risks, Snowden says we can’t afford to hold back on embracing AI. He says: “If we do that as a company or even as a country we will be at a disadvantage”.
AI is about getting a machine to do what a human can do Michael Snowden, OneNet