What can data science do for Human Resources? What innovative and valuable HR use cases are there for advanced forms of data science such as Artificial Intelligence (AI) that won’t scare employees?

AI in HR is a largely untapped area. When it’s mentioned, people tend to think of three things: 

I’m personally not keen on productivity monitoring. But the other use cases do offer real value. 

However, they can have a whiff of Big Brother to some and need to be done in the right way if ethical and bias issues are to be overcome. I’ll address what that might look like in the forthcoming piece on the key blockers stopping AI really fulfilling its potential in HR.

Here, I’d like to show you some new 2020 use cases for AI in HR that you may not have thought of before. All are examples of what I’d call an entrepreneurial approach to AI in HR: using AI in new, smart ways to solve arising issues. The possibilities are almost endless.

Compared to the usual three use cases, all four are quicker, easier, and don’t present the same issues of bias and ethics to overcome. 

AI chatbots and digital assistants

This involves using conversational AI to help employees with a business process or to answer a question. 

An employee can enter a query in text or speak it. AI is used to understand the speech or text and the nature or intent of the question. It then uses that information to decide on the best response. 

In practice, that might mean queries like this:

And up pops a number, a chart or even an infographic giving details.

At Peak Indicators, we’ve developed a successful live pilot for a major organisation to answer exactly these kinds of queries through a digital assistant. It allows HR analysts and line managers to ask exactly these kinds of questions about their workforce. 

So real-time answers to important ad hoc questions, with no fuss.

Workforce planning

Using AI to boost employee retention can add real value, but there are tricky ethical and practical questions. Firstly, if you use AI to help predict who may be likely to leave, you need that person’s explicit permission. And second, the information is of little use unless the manager has the tools – pay rises for instance – to help retain them. 

Workforce planning is different: it is all done at a summarised level. You’re not predicting which employees are going to be leaving. You’re predicting, based on previous data, a percentage: 31% of our programmers are likely to leave next year, for instance. That prediction then allows managers to handle the situation skilfully and ahead of time. Is there a problem with pay here, or with working conditions, for example. 

In that way you can make planning decisions about the workforce – the number of people of a certain grade you need to hire, by region and so on. This kind of forecasting using machine learning is much more intelligent than, for instance, using linear spreadsheet formulas. Ethical considerations are met because you are never referencing individuals. 

Hot desking and office-space optimisation

Office space costs money. A lot of money. In the City of London, the rental cost of each desk space is around £13,000 a year. 

Simple machine learning offers help here. It can use anonymised pictures from cameras to work out which desks and spaces are being used, and which areas are cold spots. 

That in turn allows big firms to start answering questions like: do we actually need this much office space? Are hotdesking areas worthwhile? Could this underused space be repurposed, say to provide more meeting rooms? It allows firms easily to optimise the workplace and their office resources.

Anonymised pictures could also be used to help ensure robust compliance with Covid rules. An algorithm could find useful patterns in whether, say, social distancing was generally being maintained in the office. If not, it could identify trouble areas to allow the issue to be managed.

Mining employee feedback and exit forms

An even bigger cost than office space to firms – especially large ones – is employee churn. An extensive report by Oxford Economics concluded that the costs of replacing an employee ran to more than £30,000, and that was several years ago.

The key way to retain staff is to listen to what they are saying. That means staff surveys and, vitally, exit interviews. Only machine learning (ML) has the capacity to interrogate these huge data sets in really rich ways to inform decision making. ML techniques can find and understand the common themes and sentiments behind the decision about leaving the company. 

That can be done for both textual and numeric information. And it’s often in the text that the real gems of information might be. ML means you can do things like sentiment analysis – working out the general feeling and sentiment behind the scores and comments. You can establish the common themes in written feedback, gathering sentiments together into topics. By classifying and categorising the nature of the comments, they can be summarised and digested by senior management. And that is very valuable insight.


Thank you! Your subscription has been confirmed. You'll hear from us soon.