No one doubts that artificial intelligence (AI) is proving ever-more reliable at predicting what groceries you’re likely to buy, which movie you’d enjoy watching and how you’d normally answer an email. Few doubt that it has enormous potential in areas such as diagnosing illness and powering self-driving vehicles.
But in managing people? For a human resources manager, much of the appeal of AI lies in the promise that algorithms can help eliminate human error and bias, helping us to make more rational, objective decisions on questions as basic yet essential as who is the best person to hire for a given position.
Can AI do that for us now? The answer is, at present, unclear. Does that uncertainty mean people managers should wait before taking AI-centered measures? No. There are a number of steps to be taken to get your people division ready to take full advantage of the promise of artificial intelligence. Here, some of the main areas:
- Data. Machine learning requires enormous troves of data to be accurate. Most HR departments simply don’t collect enough data on the people they hire – and don’t hire – in order to feed the algorithms. That means HR managers should identify what data to track and analyze, and how they can collect it. Smaller companies that are unlikely to ever produce enough data may want to look into companies that provide industry-wide pools of data and work with them.
- Bias. Examples of biased data producing biased predictions abound. Based solely on past data, most AI systems would likely reproduce biases inherent in the workplace. Part of the process of identifying the data you want to evaluate must be also identifying the biases that may be imbedded in your data and how to correct for them.
- Privacy. Personnel data is a minefield of privacy issues. It’s key to learn about what data can be used, what permissions are required, how to ensure data is accessed securely and a host of other privacy-protection measures. That will require specific training in data protection awareness for all staff.
- Skills. AI requires a different skillset than traditional HR management. You will need to up-skill people in your department to become more data-savvy, as well as to incorporate new talent in this area. It’s not that everyone in the department will become computer scientists. But everyone should have a familiarity with data.
- Integration. HR data should be integrated within the company’s business and financial data, in order to be able to analyze the effects of HR practices on overall business performance.
- Implementation. For HR, putting AI to use is about identifying data, collecting that data and then building solutions. This may be a very different approach than other technology-related changes, and requires different methodologies to put into practice.
And while recognizing the potential of AI, it’s also important to be aware of its limitations. Approach AI as a potentially powerful tool that will help you take decisions rather than make those decisions for you. Think more in terms of an assistant rather than a replacement.
As you know, human resource decisions are about people. HR outcomes are not straightforward and are difficult to measure. Employment decisions have such serious impact on individuals and society that fairness concerns – beyond data criteria — are also factors. And for now at least, qualities such as fairness and empathy, rest firmly in the hands of humans.
Based on The Future of Management in an AI World. Franz Heukamp and Jordi Canals, eds. Palgrave Macmillan. 2019