Why a Machine Will Steal Your Job
October 17, 2018
Since the dawn of time, humans have created tools to make their jobs easier. Technology has lead to the quickest and grandest societal changes throughout human history. According to History.com, the Agricultural Revolution, catalyzed by innovations like the plow and seed press, transformed our world from one with 75% of the population working on farms in 1700 to only 2% today. These tools made farming increasingly efficient, so fewer people were needed to produce the abundance of food we now have. Economies boomed as former farmers specialized and took better and higher skilled jobs.
Technological revolutions like this have happened everywhere, not just in farming. The Industrial Revolution led to an unprecedented level of production and living standards because of machines like the steam engine and cotton gin, which were creating the same goods that humans already could, but faster and cheaper.
When human and machines work together, the results are phenomenal for everyone. Living standards rise and workers specialize, moving into jobs that require more complex thought as machines take over the redundant, labor-intensive ones. However, we are currently undergoing a revolution unlike any seen before. For the first time, we are creating machines capable of thinking and learning; machines capable of taking over the same types of jobs they forced us into.
Specialized artificial intelligence has been better than humans for years. IBM’s Watson, a supercomputer with AI capabilities, answers questions stated in plain English. Seven years ago, Watson dominated the two best Jeopardy players in the world, winning by over $20,000. Within a year, Watson learned faster than the human Jeopardy champions ever could and went on to triple its score (as seen in the photo above)! But didn’t Watson have an unfair advantage? He was hooked up to the internet, right? Actually, that’s not the case, Watson never had access to the internet, it prepared for the show by reading and learning from thousands of dictionaries, encyclopedias, and history books.
You could argue that Jeopardy and IBM’s Watson relies on memorization, and not the applied learning that makes human intellect so unique. However, becoming the best Jeopardy player ever just scratches the surface of what Watson can do. Take one of the most prestigious and intellectually challenging jobs in the world: being a doctor. Here in the US, becoming a doctor is an expensive and demanding 10+ year path. Given all of this, you might think that physicians are irreplaceable, and machines couldn’t do their job better than they can. Believe it or not, Watson can beat out humans here too, and it’s not even close.
According to IBM, when diagnosing lung cancer, oncologists have a success rate of around 50%, the same probability as a coin flip. On the other hand, after just a few years of training, Watson correctly diagnoses lung cancer 90% of the time. According to the Economist, it’s estimated that doctors would need to spend at least 160 hours on medical reading per week to keep up with the cutting edge of research. Obviously, this isn’t possible for people like us, nobody has that much time. However, for Watson, this type of reading is all it does. Watson never sleeps, never takes sick days, and never asks to be paid. When all the advantages of using a machine are taken into account, it’s easy to see why, according to Forbes, AI spending will reach $50 billion by 2020.
What about creativity, isn’t that something that’s unique to humans? It’s tempting to think that the creation of paintings, music, films, and other arts certainly can’t be simulated by a computer. However, machines have tried and succeeded. According to Recode.com, the first AI-produced piece will be auctioned off later this year, and it’s projected to sell for over $10,000. In the world of music, a Princeton undergraduate coded an algorithm, DeepJazz, in less than two days. According to their website, the algorithm can produce endless jazz music for free that is indistinguishable from human-composed jazz when put to a blind test. Any skill that can be learned is a skill that a machine can do.
If machines do exactly what we do but better, faster, and cheaper, why wouldn’t companies look to replace human workers? You might be too young to work now, but you’ll have to face this problem someday. Machines like Watson can already best humans in their most complicated jobs, so is it just a matter of time before machines take every job, even yours?