Why AI is going to change the world – the Alphr view
Artificial intelligence. The words evoke images of sterile science-fiction environments, sleek robots and future-gazing articles from the past 20 years promising that, this time, AI is really taking over.
The change has been more subtle than that – not a sudden revolution where we’re servile to robots, but an undeniable shift in how AI is increasingly embedded in our technology, altering our lives as a result.
This month, alongside our other regular coverage, we’ll be paying special attention to AI with a number of interviews, features and other stories about this most exciting of technological advancements. Below, three of our writers give their thoughts on why it’s so damned important.
Ian Betteridge: Humans are right to be worried
Way back in the 1990s, I spent a great deal of time thinking about AI. Somehow, I had wangled a grant to write about it – and boy, did I write a lot of words. About 80,000, in fact, and I can still remember most of them (three years is a long time to work on a single project).
Then, AI was mostly theoretical. Now, it’s the most commonly used buzzword in tech. But is it really here?
The answer, of course, depends on your definition of intelligence. We don’t really have general-purpose artificially intelligent machines capable of passing the Turing test, for example. But what we do have is increasingly adaptable algorithms capable of learning from a set of inputs and making decisions based on experience, rather than pre-programming.
This means we’ll have machines that effectively programme themselves – and in turn, this means that we’ll no longer be able to understand their programming.
If that scares you, you’re probably right to be scared, for two reasons. First, it makes systems virtually impossible to debug. Debugging a machine that learns from its environment is as hard as debugging a human being – and debugging humans, which we usually call “psychiatry”, is really hard.
But the second reason to be afraid of AI is more of a complex issue: although we think of machines as dispassionate, every programme contains a set of hidden assumptions that its programmers have made about how things work. In the case of machine learning, these hidden assumptions, built into the initial code, can spiral into a whole nest of unintended consequences, all hidden under layers of learned behaviour.
Having said all that, I still think AI and machine learning will be ultimately be a hugely positive thing for human beings. Ultimately, where we’ll get to will be something like Iain M Banks’ Culture novels – machines smarter than ourselves, and probably with more robust ethical behaviour too. Bring on the robots.
Thomas McMullan: Apocalyptic headlines mask real societal changes
AI is one of those terms that gets thrown like wet spaghetti against every conceivable surface. It’s not always clear what people talk about when they talk about AI, but it’s flung all the same, sticking to areas like smart cities, video games, robotics, digital assistants, teledildonics – you name it, there’s wet spaghetti on it.
This sense of incomprehension only adds to the feeling that we’re woefully unprepared for what the advent of widespread, advanced AI means for our society. From government committee warnings that artificial intelligence stands to “fundamentally reshape” the way people live and work, to advances of AI intelligence in warfare, there is a vague feeling of dread about AI, coupled with bafflement about what it all actually means.
No wonder we grope for science-fiction tropes – falling on apocalyptic histrionics to distance ourselves from very real societal changes.
The good news is that it’s not all doom and gloom. The use of artificial intelligence has scope to bring many positives, particularly within areas such as medical research. Artists are playing with AI to make interesting, and filthy, neural-network projects. AI is also a security necessity, especially when you have a multitude of data points in interconnected home devices, or delivery drones flying above our heads.
This month I’ll be looking at a company that’s developing AI and robotics technologies for built environments, using large-scale 3D printing. Could AI help to solve the housing crisis? Or does it do little more than throw spaghetti on the problem?
Alan Martin: A long way to go on an uncertain path
As news editor on Alphr, I spend much of my time hearing about AI. It’s mostly in a way that consumers would never notice, invisibly making our lives a little bit easier every day. It’s silent, but it’s there and getting smarter.
I haven’t yet made up my mind as to whether I believe fears of mass job losses and apocalypse are overblown, but I’m confident that the experts who have contributed to this month’s’ features will offer insight without hyperbole.
For me, the most interesting aspect of artificial intelligence is its ability to spot things that humans remain oblivious to, such as the tiny difference between a benign seborrheic keratosis and a malignant carcinoma. This has its limits, however, in part because we humans are imperfect teachers with our own biases and mistakes that can be easily imprinted on even the blankest of canvases.
This is best illustrated by a story I once heard about an AI that was learning to recognise tanks from military photographs – it baffled researchers by correctly identifying tanks in all of the photographs – including one where the tank was invisible, obscured by a tree. The scientists scanned the picture pixel by pixel in a fruitless search for any splash of metal. Nothing.
Had the AI developed superhuman abilities? No: it had merely learned to recognise the shadows cast at the time of day when the tank photoshoot was taking place.
I like this story, because it represents AI at its most ingenious and its most flawed. The potential is undeniable – what we do with it will change the destiny of humanity for ever.