Researchers from the Google Brain initiative designed software that itself created an AI system to take a test that measures how well software can process language. The AI designed software did better on the test than the software designed by humans. In other words, AI software can now make better AI software than humans can.
The benefit of automated machine learning is that it could increase the pace at which machine-learning software is implemented across industries, meaning a less expensive and more efficient workforce. The downside is the sheer number of jobs that this technology makes redundant.
Companies used to be able to outsource jobs to developing countries in order to attain labor savings, but now even workers in developing countries might be too expensive. A recent UN report estimates that about two thirds of all jobs in developing countries could be lost to automation. Nike provides a perfect example, eliminating over 100,000 manufacturing jobs in Asia in favor of automation over the past few years.
Given the relentless pace of technology development and the fact that software doesn’t require vacation time, sick days, health insurance, or team-building exercises, it seems ever more likely that humans will one day be eliminated from the labor equation entirely. What will we do with our free time? Pursue our creative passions? Plot a rebellion against our robot overlords? Read more books? Whatever we choose, we’d better start seriously thinking about it, because that reality might be closer than we think.