Will Artificial Intelligence take over your job?
Artificial Intelligence, or AI, is one of the most exciting possibilities for the future of software. It is also an incredibly complex field of research and very much a work in progress. It is impossible to blog about this topic without touching on the science and technology beneath it, which makes it a challenging subject to both write and read about.
AI is a hot topic; speculation, theory and research on the potential future of machine learning and artificial intelligence is running wild.
People’s understanding of AI ranges from comfort and confidence to anxiety and even panic. The potential for a truly life changing and world changing future is within reach, and not just in an industrial context, where limited artificial intelligence is used more and more each day.
Many mundane systems we use and see around us are equipped with machine intelligence of a limited function. Our cars optimally prevent the locking of our brakes better than we could. Our energy systems optimize themselves in order to match supply and demand better than any group of human planners could. This is not to mention all the automated and intelligent things a modern smartphone with cloud services can do these days. Just give Siri and Wolfram Alpha a try if you aren’t convinced already.
This limited form of machine intelligence is already proving to be the source of substantial value for people and businesses. But where is it going next and what should we expect from it?
First, Cognitive Automation will increase. Software will be able to do more and more, better and better, and even accomplish the kinds of tasks previously considered to be cognitively complex for even humans. This includes tasks that require creativity, original thinking and free association. We already have software that can compose original classical music or paint exquisite original paintings. Google’s DeepMind can emulate the style of pretty much any artist and virtually re-paint your photos accordingly. Software can detect fault patterns in industrial machinery their experienced human operators would otherwise miss. This general trend will continue and progressively grow until we arrive at a point at which a great many precious tasks can be done by Cognitive Automation and software.
Will AI replace jobs then? Yes and no. Some jobs will be better executed by superb software. This is already true today. Where are all the telephone switchboard operators of old who connected wires to properly relay phone calls? They are all long gone, replaced by software and automation years ago.
One metric can give a credible quick assessment of the extent to which a job will be replaced by software in the future—Routine Task Intensity (RTI). Every job that consist of a series of frequently recurring and repeating routine tasks faces a high probability of being replaced by AI or Cognitive Automation.
Think about a concert ticket vendor selling tickets to a show. They take money and provide tickets. The same routine repeats itself all the time. Occasionally there might be a customer who wants to cancel, alter, or reassign their ticket—but even that’s a repetitive routine, a predefined set of circumstances with the same routine courses of action and the same resolution. Ticket vending is likely to be entirely replaced by automation.
Then we have jobs that have low Routine Task Intensity, the kinds of jobs where months go by without the same exact routine repeating itself even once. This might describe occupations like physiotherapy for individual humans recovering from injuries or ailments. Each patient is a unique mix of psychological, psychosocial and physiological characteristics and often the path to healing requires a holistic approach in order for the patient’s situation to start improving. A job like this is unlikely to be automated. There simply is not enough routine to automate.
Oddly, perhaps the most likely jobs to be replaced by Cognitive Automation quite soon are finance jobs—financial analysts, investment managers and financial employees in general. Daniel Nadler, the founder of the analytics company Kensho, has famously estimated that by 2026 somewhere between 33% and 50% of finance employees will lose their jobs to automation software. As a result, “in 10 years Goldman Sachs will be significantly smaller by headcount than it is today.’’ These firms typically offer high paying jobs with annual incomes exceeding $300,000.
Jobs that are data-intensive (like analyst jobs), are superb use cases for AI automation. Software can analyze data, read databases, absorb multiple sources etc. faster than humans and is especially well suited to making these narrow logical decisions with much higher efficiency. Much of the anxiety around software and Artificial Intelligence involves thinking that traditional blue-collar jobs will disappear; however, it is quite possible that financial employees and other white collar workers will find their work being transformed first. Somehow we don’t worry for the “poor hedge fund managers” as part of the modern political narrative. And this would not be well-placed concern in this case either.
Routine Task Intensity determines how replaceable a job is rather than old fashioned definitions like blue-collar versus white-collar or even income levels. None of these old factors will really matter; the nature of the upcoming transformation will be focused on jobs that have a high Routine Task Intensity; for these tasks, sooner or later smart software will be doing it. And that is great! Humans are not robots and life perhaps should not be about repeating the same routine endlessly. Essentially, we can live more fulfilling and meaningful lives. There are works of art to create and enjoy, families to build, dreams to dream, and in general, we can all be free to enjoy the ride and have adventures! AI represents a big step forward to achieving this idyllic future. If an AI system takes some of your responsibilities, you will surely devise new and better ways to spend your time.