What can AI mean to HR? Time for a reality check

The potential of AI – machine learning – in the area of human resources seems huge. The models, algorithms, tools and computing power are there. Yet are we ready to let AI take over key HR responsibilities, such as recruiting, HR administration and employee engagement? What are the quick wins and what are the hurdles to consider? Here are the findings of some experiments we have recently conducted.

Some of our AI experiments use technologies like machine learning, conversational interfaces and virtual personal assistants. We haven’t looked at deep learning, brain computer interfaces, smart robots or other experimental technologies yet, as we think it will take a few more years until these technologies are ‘smart’ enough to be used productively in HR.

One of the most popular – and successful – uses of AI for HR these days is the recruitment process. Artificial intelligence can free recruiters from the time-consuming task of sifting through terabytes of data that make up resumes, social media accounts, reference letters and other sources in order to filter the best candidates. More than that, artificial intelligence can help recruiters predict the match between a candidate and the open position using questionnaires or even public professional profiles. It can even predict employee retention rates – with accuracy percentages of over 90%.

AI applications like the ones described above, however, require loads of data. Having data on a population of 1,000 people over a five-year period probably isn’t enough to train a machine-learning model to come to significant conclusions. On top of that, the data must be high-quality and coming from diverse sources – more than just your HR information systems. It’s only by combining these data with social media data, email traffic, sentiment analysis, external data, demographics, etc. that you’ll get valuable insights.

When handling personal data, however, it’s important to ensure that you’re not crossing the line in terms of what is permissible under privacy legislation (GDPR) and business ethics. Data, and HR data in particular, should be treated anonymously and confidentially at all times.

Since delaware started its journey with intelligent chatbots, we’ve become fans. AI assistants can be irreplaceable helpers to businesses. Again here, the possibilities for HR are nearly endless. Within an HR department, more than half of all inquiries received are routine questions about sick leave, holidays, payments, etc. It’s perfectly feasible for a chatbot application to answer these routine administrative questions and perform simple transactions.

As an example, let’s say an employee wants to inform their employer that they are ill. Normally, they would go the doctor, get a sick note and and submit a report through the company’s HRIS application. Wouldn’t it be easier if the company’s chatbot could inform the manager, activate the ‘out-of-office’ message and upload the sick note to the right place?

The application gets tougher, though, if the employee wants to report that they don’t feel ill anymore and want to return to work earlier than expected. In this case, the bot needs a fair amount of cognitive intelligence to ‘understand’ what the employee ‘intends’ to do. Training a bot in detecting intent correctly is absolutely crucial for a successful chatbot project.

Chances are high that your employees are still very skeptical about talking to a machine. That is why it is key to make sure the chatbot application you’re using proves its worth from day one. Explaining to your users that it IS a bot and not a human will help, too. Last but not least, make sure that there’s a backup: having the possibility of forwarding a chat session to a ‘real live’ expert is a very valid scenario.

Introducing chatbots is, therefore, a joint effort of the IT and HR teams. Asking the IT guys to build the bot themselves without considering user interaction (the conversational design) will not lead to success. In essence, the technology is not that complex… but what your employees expect is an easy, fun and efficient interaction.

In summary, AI for talent holds great promise and much is already possible today. Machine learning can help the HR team predict who is the best candidate for an open position, who is at risk of leaving your company or who will become an outperformer. To make it work, however, data is crucial, as is open communication and a human ‘fallback’ scenario. After all, in HR – just like in all other domains – the smartest deployment of AI is one that augments human expertise rather than replaces it.

Get inspired by the opportunities of AI for HR: reach out to our experts for more information.