Conversational AI: Is It A Conversation?

Tom Bennett

A colleague of mine recently referred to “conversational AI” in the context of human resources services while speaking with a customer. Just in case you aren’t sure what conversational AI is, it’s any machine that a person can talk to, including chatbots on a website or social messaging app, voice assistant or voice-enabled devices, or any other interactive messaging-enabled interface. People can use these kinds of services to ask questions, get opinions or recommendations, find support, complete transactions, or have virtually any other interaction where computer based-technology is present.

There are some easy-to-understand use cases; recruiting and career sites, for example, where a chatbot can help job applicants search for suitable positions, apply for a job, and sign up for career updates. Another common example is asking a “smart speaker” to play a certain song or answer a question (“why is the sky blue?”). The idea is to provide a way for humans to interact with technology that is natural but also helpful. Conversational AI and digital assistants are set to take a prominent place in the near future with regard to HR interactions and a host of other areas, potentially lightening the burden on HR professionals and providing a superior customer/employee experience. But is this technology really engendering a “conversation”?

This got me thinking about the place that such technology has in the realm of HR conversations. Nearly half of adults use digital assistants every day, and 87% say it is more convenient than speaking to a human, according to the Pew Research Center. Technology has long been a key part of HR management. What possible advantages could such technology provide above and beyond actual humans, for example? Why not just staff up with low-cost human labor resources?

It turns out that there are a host of benefits available when artificial intelligence-enabled agents are used for specific scenarios. Call me a skeptic, but until recently, I wasn’t convinced that this technology really provided any advantages over … well, humans.

The most prevalent justification for conversational AI technology is where it can supplement processes that are deemed too mundane, tedious, or just plain boring for humans to do. Conversational AI can be used where organizations simply don’t have or cannot afford enough staff to provide efficient service to their customers, whether customers are actual purchasers or, in other scenarios, job applicants, or perhaps employees seeking policy information.

It seems fairly evident now that some of the benefits are around quick answers to common questions, which can provide a much more satisfying experience for the recipient of the conversational AI services. This technology is ideal, for example, at collecting resumes, extracting the details from the resume, and finding matching jobs to that person’s skills and backgrounds. If you have ever tried applying for a job on a career site, you know the process can be tedious. AI-based assistants like chatbots can make the process easy, simple, and efficient. It’s helpful, therefore, to understand what kinds of conversational AI technologies are available to support these kinds of bot-human interactions.

Most prominent are chatbots, which as the name suggests, interact with people as part of a chat session. In this case, typing is a key part of the interaction. If you aren’t proficient at typing, the interaction could be frustrating. This brings us to another type of conversational AI: voice assistants. You have no doubt seen or interacted with Apple’s Siri, Google’s Assistant, or Amazon’s Alexa agents. With voice assistants, the user speaks out loud and the AI responds in spoken form. Lastly, similar to voice assistants are mobile assistants, which use the same voice assistant and chatbot technology on mobile devices like smartphones.

woman interacts with a digital assistant device in her home while washing dishes

Anyone who has used voice assistants knows that, at this stage in their development, the results can be sporadic and unpredictable. Even where there is 80% accuracy, the interactions can be quite frustrating, if not humorous. For this reason, chatbots can be a more attractive option, as a request from the human typing can be interpreted more accurately.

If you are a decision-maker in an organization considering conversational AI, what is the right choice? I think this depends on the application, the brand, and what you expect to deliver to your customers. Trying to provide technical support at this stage in the technology’s development, for example, is probably not feasible. The terms used, the frustration inherent in the troubleshooting process, and many other factors are likely to lead to a bad experience for the customer. However, leading a customer to the right solution article, department, or an actual person is already working well. Providing basic information like “how much am I paid?” or “how can I apply to take leave?” can effectively be done using conversational AI. What will matter going forward is our relationship to these AI tools.

A recent article in the MIT Technology Review described how Saint Louis University provided all its dorm-dwelling students an Amazon voice assistant. The study outlined some of the many benefits from a human resources perspective, like knowing when the dining halls are open or details about campus activities, or even just providing the feeling of companionship for new students.

But there is some doubt about having all that data and a recording of the human interactions of students kept by Amazon. What will Amazon do with that data? Is it private or anonymous? Many of my friends and colleagues have asked the same questions and won’t allow such a device in their homes for fear of spying. Despite these misgivings, to a certain extent, the steady march forward of having intelligent assistants listening to us all day every day seems inevitable, and many organizations and countries are considering laws to regulate how this information can be used.

call center employees talk to customers through their headsets

Conversational AI is an intriguing technology, and one that is constantly evolving. Its place in the world of human resources is still uncertain, and it may turn out that the best solution in many cases is still a human being. However, the benefits from an automation and efficiency standpoint are undeniable.

The biggest challenge, in my opinion, is getting humans – customers, users – to trust that the technology really will help them and not just provide a frustrating experience. Does this chatbot or voice assistant really care if my problem is solved; is it invested in my wellbeing? Obviously not. There isn’t a “real” conversation going on there, just an algorithm “listening” for keywords. For this reason, there will be limitations in terms of what is acceptable for AI to manage, especially when it comes to brand management, perception, or sensitive issues. Can you imagine reporting a bullying incident to a chatbot?

Conversational AI is evolving every day and will have a prominent place in the future of human resources, perhaps reaching the lofty goal of passing the Turing test and becoming so reliable we can forget we are talking to a machine. I, for one, welcome our new AI overlords … I mean conversation partners (gulp!).

Get up to speed with Accelerating Customer Experience With Conversational AI.

Tom Bennett

About Tom Bennett

Tom Bennett is a senior solutions consultant, focused on the Asia-Pacific region, at SAP. Drawing from more than 25 years of HR experience across multiple industries and countries, he evangelizes the power of HR software and services from SAP. As a dual citizen, Tom is based in Perth, Australia, but is originally from Los Angeles, California. He is currently working with mining and resources companies to help them understand the art of the possible for enterprise software and processes.