I’ve been looking forward to writing this post. Since this HR technology series has started, we’ve talked about enterprise resource planning (ERP), application program interface (API), cloud computing (aka “the cloud”), and many more. Bots (also known as internet bots or web robots) have become mainstream vernacular. But what are they really and why should human resource professionals pay attention to them?
To help us understand more about bots, I asked Deepak R. Bharadwaj, vice president and general manager for ServiceNow’s HR Service Delivery product line, if he would share his expertise. Prior to ServiceNow, Bharadwaj was vice president of product management for Oracle’s talent acquisition and talent management solutions. He’s also held leadership positions at Taleo and SAP. He holds a master’s degree in computer science from Clemson and graduated with honors from the MBA program at the Wharton School. And I’m thrilled he agreed to this interview.
Deepak, let’s start with a definition. What’s a bot and how does it work?
[Bharadwaj] If you do a search for the term “bot” you will quickly find that bots are applications that perform repetitive automated tasks at a scale that is not humanly possible. Over the past two decades, marketers (and unfortunately spammers) have successfully used bots to reach many individuals simultaneously to accomplish their goals. Because of the advances in artificial intelligence (AI), we’re seeing a positive resurgence of bots.
Bots have become synonymous with chatbots – applications that can understand what you are trying to ask and then respond with the right answer just like a human would. The holy grail is for these bots to exhibit intelligent behavior that is virtually indistinguishable from that of a human.
How could a bot help human resources departments?
[Bharadwaj] Just like they do in customer service, chatbots have immense potential in providing service to employees across all departments including IT, HR, legal, marketing, and finance. Specifically within HR, there are typically three major types of conversations that a chatbot is likely to have:
- Retrieving system of record data such as “How many vacation days do I have left?”
- Answering a question based on knowledge base such as “What is the leave of absence policy in California?”
- Submitting a transaction such as creating a leave request and getting updates on approval status.
HR business partners and HR departments today are spending enormous amounts of time answering basic questions and fielding tactical requests from employees. A range of research studies from McKinsey to the HR Trend Institute put this number at 60-70 percent of the time.
HR departments have typically addressed these challenges by setting up service delivery models that can deliver service easier and faster for less cost. One of the key elements of these service models is case deflection – or the ability for employees to find answers to questions and address any needs themselves without having to go to HR. The primary case deflection approach today is to search for answers in what is typically known as knowledge base.
However, chatbots can help HR departments provide a more modern conversational experience to getting personalized answers and solutions thereby dramatically improving the case deflection rate and reducing the workload on what are typically known as Tier 1 and Tier 2 – the frontline HR support folks. Thus, HR can scale to support their employee base much more efficiently and effectively and truly free up personnel to work on more strategic initiatives.
Are chatbots a component in software or do they stand-alone?
[Bharadwaj] A chatbot needs to understand what you are trying to ask of it. This is typically accomplished by technologies that are a combination of conversational design, pattern recognition and natural language processing. IBM’s Watson and Google’s API.ai are two well-known examples of platforms that provide publicly available conversation services for chatbot applications. Conversation services can be stand-alone or included as a component in HR applications or a combination of the two. The conversation design itself may be part of the HR application, but the application may rely on third parties like Google and IBM for natural language processing services.
Once a chatbot knows what the user is trying to accomplish, it must execute the conversation. Sometimes it can be straightforward if you only want to retrieve data. The system knows who you are and easily gathers data related to your profile. It gets more complicated if you need to submit a transaction. The chatbot now needs to ask for a few pieces of information that go into the transaction and these data inputs will often vary depending upon the use case.
For example, a request for long-term disability leave may not need end data, but one for a short-term disability request might. Because the logic of what data fields to gather for different requests is deeply embedded in the application, chatbots performing these tasks are better off embedded in the application itself.
When answering questions based on knowledge base, chatbots must be part of the application that contains the knowledge base. These chatbots are the most complex because they must comb through knowledge bases and present answers that are relevant and personalized based on the user’s role, job type, location and a host of other criteria. While it is still early days for this type of technology, there is a tremendous opportunity for designing knowledge bases in an AI-first world.
Are there any downsides or challenges that HR pros need to be aware of when it comes to chatbots?
[Bharadwaj] HR departments must recognize the effort it takes to identify the types of conversations they would like a chatbot to have and then to create those conversations. One key question that HR teams often debate is whether to be transparent to employees that they are actually having a conversation with a chatbot and not a human.
Chatbots may not be the best paradigm for all interaction with systems. If you need to browse a catalog and visually inspect items before making a choice, a rich media interface works much better than a chatbot. HR must invest in service design and employee experience design skills if they want to offer employees an experience that integrates chatbots.
Last question: what do you see as the future of chatbots for HR?
[Bharadwaj] Most chatbots today have limited AI capabilities. They can either have a conversation on a programmed topic or they will bring in a person if they don’t know how to have a conversation. Expect next generation chatbots to learn on the fly. Every time a person needs to be brought in, a chatbot might ‘listen’ to how the human has a conversation and program itself automatically to have that conversation the next time it is called upon. Intelligent chatbots have immense potential in shaping the way we interact with systems in the future and this is just the beginning.
I want to thank Deepak for sharing his expertise with us. I know this is a lot to digest and it’s only the beginning of what chatbots can bring to our organizations. But chatbots aren’t a fad and they’re not going away. Even if we don’t have a chatbot, we need to start thinking about them. And we’ve inspired you to learn a little more about technology and HR, you can check out the rest of our series.
Image captured by Sharlyn Lauby at the National Atomic Testing Museum in Las Vegas, NV3