Just because a machine can learn from mistakes doesn’t mean it is self-aware and about to deploy robots to destroy humanity throughout time and space. But it does mean that increasingly, machines can take on more and more human work.
On 11 February this year, President Trump signed an executive order directing US government agencies to prioritise investments in Artificial Intelligence (AI) research and development. There isn’t any detail on how the AI Leadership executive order will be paid for, but as a statement of intent right from the top, it’s pretty powerful. So, is this something you need to worry about? Will robots be taking your job next Tuesday? Probably not, but the answer is not as reassuring as it sounds.
When we think of AI, we probably think of Skynet (the evil computer that hunts humans in the Terminator films) or the similar tricked-up calculator that is the meanie in the Matrix films. But real AI is a little more mundane. It is more likely to be making sure your car headlights are on when you need them (and not on when you don’t), sending a nuisance spam call to your voice-mail or suggesting the next thing to watch on Netflix. AI is the catchall term for software that can solve problems based on rules rather than a linear set of fixed instructions. Really advanced AI can modify the rules based on how things turned out the last time or patterns that it detects in the environment.
Just because a machine can learn from mistakes doesn’t mean it is self-aware and about to deploy robots to destroy humanity throughout time and space. But it does mean that increasingly, machines can take on more and more human work. In recent decades we have seen this kind of automation steadily eat away at assembly line jobs as increasingly AI driven robots replace workers performing limited and repetitive functions. A robot can sort big apples from small oranges more efficiently than a human and it never needs to take a break (or be paid).
As the technology advances, it’s starting to creep into areas we might have thought of as immune from automation. Medical diagnosis is increasingly the target for deep learning AI, the kind that recognises patterns and makes predictions based on those patterns. During their career a doctor might see a few thousand x-rays or MRI images and get better at noticing patterns. But AI software can review every x-ray ever made before the doctor has finished her morning coffee.
A recent study, for example, compared the diagnostic precision of AI software with that of teams of specialist doctors from all over China. The AI software was 87 per cent accurate in diagnosing brain tumours in 15 minutes. The doctors could only diagnose 67 per cent and needed twice as much time to do it. The AI increased precision and saved time because it was able to learn from a much larger base of experience than any individual doctor or team of doctors ever could. It uses like this that are why AI is predicted to add $15 trillion to the global economy by 2030.
President Trump joined the 18 other countries that have announced AI strategies since March 2017, because he wants the US to be a leader in AI rather than a follower. And it is why investment in AI based startups jumped 72 per cent to almost $10 billion in 2018 alone.
And even though some analysts are predicting 1.8 million jobs will be lost to AI in 2019 alone, those same analysts are predicting that the AI industry will create 2.3 million jobs in the same timeframe. You can’t buy buggy whips now because the industry that created them was destroyed by Henry Ford, but there are many more jobs in the automobile industry he created than there ever were in the one he killed.
When analysts from McKinsey looked at the employment impact of AI in five sectors last year, they concluded that jobs which use basic cognitive skills, such as data input, manipulation and processing will likely decline, while demand for higher cognitive, social and emotional, and advanced technological skills should grow, as will the number of jobs that require customer and staff interaction and management.
If your job could be classified as administrative support then the future does not look bright. And even if it requires you to do years of training so you can manipulate or recognise patterns in data, like those Chinese doctors, a financial analyst or a military strategist then AI will be coming to a workstation near you within the foreseeable future. Humans are still a little too messy and unpredictable for the average AI bot. So, if your job needs you to interact with humans and please them, such as in direct sales, management or counselling, then you are probably safe, for now. And of course, if you are writing the programs that drive the AI then your career is assured.
AI is rapidly changing the face of the modern workplace. And while nothing much will change by the end of the year, by the end of the decade, most jobs will be unrecognisable. You’ve been warned. It’s time to transform yourself from a data geek to a people-person, before your computer takes your job.