“Machines Learn and develop Opinions. The Machine Recommends. Humans must make the decisions while the organization inherits the liability. Saying that ‘the computer made me do it’ won’t be a workable defense. Make sure your vendor discloses every aspect of their algorithms and test to see if the systems recommend your people.” - John Sumser

Three things all companies should be doing with their HR Tech
to prepare for their future with AI and Intelligent Tools.

Despite all the talk, we are still in the very early stages of the introduction of AI and related tools to the worlds of HR and Recruiting. It’s a lot like the early days of the airplane or the automobile. There were about 80 companies investigating flight when the Wright Brothers first flew. It took a very long time for their triumph to morph into useful aviation. There were lots of opinions about what would, could, or should work.

2017-04-21 HRExaminer photo img sumser john bio pic IMG 3046 black and white full 200px.jpg

John Sumser is a Principal Analyst for HRExaminer.

Three things all companies should be doing to prepare for their future with AI.

  • Training: Teach employees to argue with the machine. Machine intelligence produces probable insights. Like Vegas odds, they are not a guarantee. The best thing you can do is start by assuming that the recommendations are wrong.
    • Ethics: Build an ethics board that includes people outside HR and the company. Ask – What could go wrong?
    • Legal: Never put an algorithm in play without a review by your company lawyer. You need to know where the intelligence is in your software. Anytime a machine gives decision making input, it should be vetted for bias, appropriateness, compliance with policy, and unintended consequences. Since the employer is liable for the results of the recommendation, it’s best to keep the lawyers involved.

Here’s a deeper dive into those three topics:

  • Machines Learn. That means that their performance will be variable. Work with the vendor to establish a reliable way to monitor the variations.
  • Machines have Opinions. Humans make decisions. Be prepared to trust your gut even though the device holds the data.
  • The Machine Recommends. The organization inherits the liability. Saying that ‘the computer made me do it’ won’t be a workable defense.
  • Make sure the vendor discloses every aspect that has an algorithm or model. For instance, automated interview scheduling tools will perpetuate the bias in the hiring system. Meanwhile, the underlying algorithm or the fact that machine learning is involved may not be apparent to you. Always get a disclosure describing every place where intelligence occurs to the process.
  • Test to see if the systems recommend your people. Finally, the only way you can be sure that the machine is doing what you want it to do is by calibrating with test data. One good way to do this in HR and Recruiting is to make sure that the system always recommends your current team first. Get their data into the system and use it as the primary benchmark.

Read previous post:
Sacred Cows Make Great Barbeque

The idea is so fascinating it became the inspiration behind one of my most popular original quotes, “Sacred cows make...