Many people will soon lose their jobs to robotics and artificial intelligence (AI). That’s no longer a question but an inevitability. And those mostly in danger of being replaced are manufacturing and production jobs. No one ever thought the job of programmers might one day be endangered by AI. Yet that’s exactly the predicament we’re facing right now.
According to Google Brain AI researchers, they have designed a machine learning system that’s capable of designing and developing machine learning software. And apparently, the resulting software that came from machines coding other machines was similar or sometimes even better than those done by humans.
Wherever this development takes us, one thing is clear — machine learning experts who are currently earning premium pay because of their specialized skills and limited numbers may now have to entertain the prospect that their jobs are not as safe as they thought. While it isn’t likely that they can be replaced anytime soon, the possibility is there, and they have to be prepared for it.
Meantime, there’s another side that has to be considered. As Jeff Dean, Google Brain research group leader, says, machines that can create machines will be helpful in easing the workload of current machine learning experts. This, in turn, can accelerate development of AI software so its applications can be more extensive, far-reaching and practical.
For their experiment, the researchers challenged their software to create machine learning systems — what they are referring to as systems that are ‘learning to learn’ — for several kinds of different but related problems, like navigating through mazes. And what their software turned up were designs that demonstrated an ability to generalize and pick up new tasks with less additional training than what was typically needed. They believe these types of systems will lessen the requirement for large amounts of data currently being used by machine learning software to perform tasks efficiently.
During the AI Frontiers conference held in Santa Clara Convention Center, California several days ago, Dean mentioned “automated machine learning” as the most promising research avenue for his team. He was also quoted as saying: “Currently the way you solve problems is you have expertise and data and computation. Can we eliminate the need for a lot of machine-learning expertise?”
If we didn’t know about their research, we might have simply thought that Dean was asking a rhetorical question. But we know better. His question is more like a glimpse into what they are working towards achieving.
Obviouslyi it’s an ambitious goal, and it’s hard, at least for the time being, to decide if we really want the team to accomplish what they want. As reported in MIT Technology Review, Otkrist Gupta, a researcher at the MIT Media Lab, said: “Easing the burden on the data scientist is a big payoff. It could make you more productive, make you better models, and make you free to explore higher-level ideas.”
But at what cost? If Jeff’s team succeeds, more human workers will obviously be displaced. And we’re no longer just talking about mechanical jobs, or the ones with repetitive tasks. In time, even the jobs that require creative thinking will eventually be taken over by AI.
Nobody likes being replaced but the reality is that the age of intelligent machines has arrived. And with the way AI is developing, it’s becoming harder not to feel threatened by the possibility that at some point in the not too distant future we will unavoidably become replaceable. The question is: will we ever be ready to objectively face that reality?