Some people are excited about artificial intelligence (AI). In human resources, proponents talk about a “new era” of higher-quality outcomes, tools for innovation, and enhanced employee and jobseeker experiences. Some claim that AI-powered HR chatboxes empower employees with fast answers and self-service support. Is it all true?
There are success stories
Google’s Project Oxygen used AI to identify the characteristics of effective managers within the organization so that specific training programs and feedback mechanisms could be developed to promote those characteristics among their other managers. Google’s size no doubt made this more feasible.
AI is extensively used in recruitment. Johnson & Johnson had it identify unconscious bias in their job listings and scored a 9% uptick in female applicants by eliminating language that showed a masculine tilt. Many companies also use AI to screen résumés, boosting productivity.
And there are not-so-successful stories
In a survey of more than 2,250 business leaders in the US, UK, and Germany by Harvard Business School professor Joe Fuller, 88% of executives said that they knew their AI recruitment tools were rejecting qualified applicants. Algorithms may weed out candidates because of perfectly explainable gaps on their résumés or because they are missing one or two skills on a needlessly long list.
It’s all about the data
It’s important to remember that AI is data driven. It mines vast databases in search of meaningful patterns it can apply to real-world decision making. That means things can go wrong if the data is not ideal. Amazon trained its résumé screener tool on the résumés of existing staff, who skewed heavily male, only to discover that the tool began downgrading applicants with the word “women” on their résumés, such as “women’s soccer team.” When a fix didn’t work, the tool was discontinued.
Employee privacy is an issue
Another issue is privacy. Much of the data in HR is personal. Before creating or rolling out systems that collect and use such data, it’s important to make sure it’s secure and that you have addressed privacy concerns. You need to let people know what you’re doing and give them a say in how you do it.
While it has drawbacks, AI has great potential. It can turbocharge productivity and accomplish things that were unthinkable only a few short years ago. But it’s a tool, and like any tool, it has to be used with the discernment that only humans can bring to the table.
