September 2014 | Riviera Partners
Machine learning has gone mainstream. It started with text and character recognition used by the USPS and has led to voice language processing with the likes of Siri, biosurveillance by the world Health Organization and microtargeted marketing from just about every company you can think...
Machine learning has gone mainstream. It started with text and character recognition used by the USPS and has led to voice language processing with the likes of Siri, biosurveillance by the world Health Organization and microtargeted marketing from just about every company you can think of. And as evidenced by the Google driverless car, the self-driving car may not be far on the horizon.
But machine learning has an image problem, because what happens when the machine breaks? And it’s not a question of if–it is going to happen at some point. Case in point: In 2011, an unassuming scientific book popped up a $23.6 million list price on Amazon due to a robot price war caused by unruly algorithms. In a manner of speaking, machines are human too, in that they make mistakes.
Now take self-navigating cars. Once they hit the road, there are still going to be accidents. While the probability is likely to be much lower than with human drivers, who crash with alarming frequency, people are much less likely to forgive a robot. Machines learning methods are also 15- to 25-percent more accurate at predicting cancer susceptibility, recurrence and mortality, but we expect better. A 60-percent success rate isn’t good enough from a machine.
The fact is–and this should be comforting to many–the human role in artificial intelligence applications is far from obsolete. The objective of machine learning in some sense is to transfer some control from person to machine, but it’s still a human + machine equation. We’re responsible for paying attention to when things are changing in the environment and putting processes in place to control glitches in the system. Humans need to contribute their unique data in order to continuously improve and innovate.
The fact is, we’re the ones with intuition and creativity, not to mention the ability to understand humor and sarcasm. And there’s no doubt we’re better at relating to other humans and solving for certain problems. When it comes to recruiting, it’s not enough to simply have access to the data and algorithms that know how to populate certain patterns. You need skilled people who not only know how to analyze those patterns, but also know how to spot gaps in the data. At that point, one-on-one human interaction might be the best way to find the best match between a client and a candidate.
Machines will continue to improve in their usefulness in many verticals, and recruiting is among them. But as long as machines remain imperfect–and there’s a need for a need for human interaction in the system–humans are far from becoming obsolete.
November 2018 | Riviera Partners
Finding a new job is almost always a strategic career move and announcing that change to your network should also be approached thoughtfully. Jodi Jefferson from our New York office shares some guidance on how to approach sharing this exciting news with your network. (AlleyWatch) How...
October 2018 | Riviera Partners
Thinking About Diverse Teams as Systems by Jodi Jefferson, Riviera Partners It’s been a busy year for diversity in the news. Since the Google Manifesto and the Uber debacle, it has become clear that even large, forward-thinking tech companies continue to struggle with diversity...
October 2018 | Riviera Partners
Machine learning is a method used to devise complex models and algorithms that lend themselves to prediction – it sounds complicated and cumbersome, but this data really tells a story. Recently one of our engineers questioned why some searches take longer than others and decided...