Todd Jochem and his wife Barb renewed their marriage vows inside the self-driving car at a drive-through chapel in Las Vegas, their friends tied tin cans to the back bumper as part of the celebrations

Luke Dormehl: 'Thinking Machines'

We live in a world where we have started to take artificial intelligence for granted. According to Luke Dormehl, there’s more to the story than simply programming machines to do our thinking for us.

The ‘race for artificial intelligence’, as the author describes his narrative in the subtitle of his new book, is for Luke Dormehl a fascinating subject: “We all feel that we understand AI to some degree because it is something that has come out of popular culture. Science-fiction writers were addressing this a long time before the technical field itself was formalised as its own discipline in 1956.”

He explains that this date is a starting point from the modern perspective: prior to that the idea stretches back to the times of Ancient Greece where there were myths of bringing life to the inanimate. “Even though we have an understanding of what the modern systems are, the way that they function, which is often fraught with interesting ethical dilemmas and technological challenges, we are unaware of how a lot of it works.”

It followed then that Dormehl’s aim in his superb ‘Thinking Machines’ was to write a history of AI, “hopefully uncovering some interesting stories. There are people like Alan Turing, and the public is familiar with him. But you’ve also got other people who were hugely influential, whose names don’t necessarily come up in shorter articles about AI”.

This perceived gap in public understanding lies between the advent of Turing’s ‘Old AI’ and the arrival of ‘New AI’, the data-driven world of Google dating back to the 1970s. “But there’s a lot of interesting stuff in the middle, as well as a certain amount of secrecy in that the systems themselves are so complex that we can’t understand them easily. I wanted to write about how these systems work in a way that was grounded in what was actually happening, rather than a dystopian far-flung view of where we will be in 50 years time.”

Dormehl says that he’s always been “something of a geek, enjoying taking computers apart and putting them back together again”. But his inspiration for writing ‘Thinking Machines’ came more from the human-interest story of the people responsible for creating the “great invention of artificial intelligence that has become such an important part of all of our lives”.

He also has what he describes as a “soft spot for the early days of AI when the rules were being written as we went along. I’ve tried to include plenty of stories from these days when the first wave of creators were building machines like Shakey, which was an early robot that could move unassisted. At the time, journalists got carried away with the story and they wrote about how it could be left on the Moon for a decade and wouldn’t need any instruction. Of course, that was horribly overblown”.

Dormehl’s favourite anecdote is that of the Carnegie Mellon University researcher Dean Pomerleau, who devised a predecessor to Google’s self-driving car in the 1990s. “I was fascinated to discover that he had built this machine using a computer that’s far less powerful than one of today’s smartphones and yet managed to drive it across the US. I think this could make a great movie, but it’s also interesting because the technology could detect the lines of the road ahead of it and steer assisted by them. This was fine until it went past a tree. But it was the start of a mainstream technology that had originally been a fun lab experiment.”

One of Dormehl’s key concerns in writing ‘Thinking Machines’ was to “move the conversation away from the doomsday scenarios you get on a five-minute segment on TV news programmes that just talk about the dangers of AI, what happens when machines are smarter than us and take over because they no longer like their creators”.

He says that there are “plenty of things to be concerned about when it comes to AI. But that area has been explored and there are far more pressing issues to address in the immediate future.”

While there is the temptation to say that in a century’s time we might reflect on how we allowed AI to get out of hand, he adds, “that tends to be true of all technology that extends our abilities beyond what is natural. Splitting the atom created the nuclear bomb, but also created the possibility of cheap unlimited energy for us”.

Likewise, historically there is the same duality of outcome with the printing press, the Internet and even bridge building. “We can’t stop creating AI and we can’t decide where to draw the line. When does the sort of AI used to create compelling battle scenes in a movie such as ‘Lord of the Rings’ become a real-life automated, remote-control weapon of war?” It’s a classic case of the idea being in conflict with application. “And if that’s the case, should we be more focused on creating even better AI in order to prevent ourselves creating artificial stupidity?”

You can’t talk about AI for long without bumping into the societal aspects of the development of the technology. “I’ve spent a lot of time thinking about this and hopefully I have a slightly different take on this to most people. It’s not just about the division of society across a technological fault-line because, increasingly, everyone is becoming more technologically aware, if only to a certain extent. If a division exists, it is between the passive user and what academic McKenzie Wark calls the vectoral class - in other words the people who control the systems. It would be very reductive to say that social changes, such as the polarisation of rich and poor, are exclusively driven by technology. As easy as it is to blame AI for issues such as this, it’s worth remembering the sustained prosperity of the post-war world was brought about by technology.”

As the conversation drifts to the ethical, philosophical and legal aspects of my intelligent lawnmower inadvertently running over Dormehl’s dog, one thing becomes clear. In many ways, although a self-contained book in its own right, ‘Thinking Machines’ is also something of a first chapter in a series of assessments that need to be addressed as the pace of evolution of technology outstrips the ability of legislation to keep up with it. While AI might well account for a reduction of jobs in the future, there will be plenty of openings for a new breed of digital ethicist. *

Thinking Machines’ by Luke Dormehl is published by Ebury, £14.99

Recent articles

Info Message

Our sites use cookies to support some functionality, and to collect anonymous user data.

Learn more about IET cookies and how to control them

Close