Thanks to decades of sci-fi, the idea of artificial intelligence comes with a lot of unrealistic baggage. But the technology is now being routinely used in the working world in the form of machine learning (ML). Machine learning is best defined as a program that is able to analyse data, learn from it, and establish patterns and connections independently without direct human intervention.
The automatic assistant
Machine learning can be a powerful aid for many decision-making processes within the HR department, such as deciding on the right benefits schemes to invest in. It can also be a useful addition to the frontline of HR support in the form of an ML-powered chatbot.
Simple chatbots have previously been used to detect keywords and provide limited answers or direct users to the right resources. This same approach can be used by the in-house HR team to help address common questions and save time for human HR staff.
However, machine learning has enabled chatbots to advance far beyond a simple question-response system capable of dealing with only basic queries. By feeding an ML-powered chatbot program all of a company’s HR policies the system can understand and contextualise them. This means it will not only be able to identify pre-designated keywords, but will also start to understand and answer real questions, even when they are worded in new and unexpected ways.
Creating context
An ML-powered chatbot’s ability to continually develop and add new context to its understanding is what sets the technology apart and elevates it above a simple FAQ alternative.
Unlike a search engine function, a machine learning program can piece together data from different sources and establish connections in an intuitive way. For example, by feeding a program information on the workforce’s leave allowance and calendars, as well as the company policy handbook, it is possible to create a system with a complete understanding of the annual leave process.
So, for example, an employee could ask if they can book next Thursday off and the system will be able to tell them what their remaining days of leave are, and if anyone on their team has conflicting annual leave booked.
This can even be taken a step further by allowing the chatbot to initiate a request for leave, letting the employee do everything in one quick conversation and all without any need to speak to the HR team.
In another example, let’s say we have an employee who needs to take a sick day, which they log with the HR system. Two days later they are still unwell and unable to return to work. The ML system detects this, and is also aware that they have a medical insurance package through the company. It can send a message to the employee asking if they would like to activate the package, and also recommend the use of any other relevant health benefits, as well as reminding them of obligations such as doctor’s notes.
From here the chatbot can then hand over to a human HR worker for more involved communication and decision-making. Even at this level of transaction the chatbot is still an augmented support agent designed to help rather than replace human HR staff.
Continued development
It’s worth noting that machine learning is still in its infancy. While we have seen very successful pilots so far we are still only scratching the surface of what the technology can do. A unique benefit of machine learning is that a program will continually learn and develop with minimal human guidance. Furthermore, once a program has been trained on a dataset it can be copied and deployed elsewhere and easily apply this context and understanding to a different company’s policies and employees.
Staff can also greatly help with the advancement of chatbots by providing feedback. Rating an engagement out of 10 and providing short comments can help the chatbot learn from its mistakes and take on helpful new traits.
As companies continue to experiment with what is possible with machine learning we will see increasingly more advanced chatbots that can take even more burden from human HR teams and free them up for higher-value activity.
James Whelan is managing director of Avantus Systems
