While chatbots have been around for decades (the first chatbot ELIZA was created by Joseph Weizenbaum in 1966), only in recent years have chatbots become a ubiquitous trend – in the form of virtual assistants, product trainers, customer service agents, and online tutors, to name a few.

By definition, a chatbot is a computer program designed to simulate conversations with humans through websites, mobile apps, wearable devices, or home appliances. In the learning and education domain, there have been a few successful cases that suggest the potential benefit of chatbot usage, particularly in settings where there are large bodies of learners to support. One such famous example is Jill Watson, the artificial intelligent (AI) teaching assistant that was based on IBM’s Watson platform, and developed to handle questions posted in a discussion forum for Georgia Institute of Technology’s online computer science program. Interestingly, none of the students in the course suspected that Jill was not human, and one student even wanted to nominate her for a best teaching assistant award!

In the corporate sector, chatbots are just starting to flourish in learning and development. The following list provides some ideas and possible applications for the use of chatbots in workplace learning.

• Reference and instructional assistance: Organizations often have information in silos, spread across many departments, and company intranets could be complicated to navigate with poor information architecture. Chatbots can simplify the hunt for information across multiple sources with few clicks, make recommendations and keep a record of the conversations. Furthermore, some chatbots can conduct content analysis to further refine and personalize the information to individual users.
• Performance support: In addition to users seeking information, chatbots can anticipate questions and interact proactively to push out performance support material, particularly for a mobile and geographically dispersed workforce. The ability to provide consistent answers is useful for compliant-based content or any other materials that require a high level of standardization. Chatbots’ 24/7 availability also work well for shift workers or for global companies with offices in multiple time zones.
• Coaching and mentoring: Not only can chatbots scale personalized coaching and mentoring sessions, they help to make learners not feel judged when they ask questions and seek clarifications. One of the challenges in coaching and mentoring is for people to feel comfortable to talk openly and honestly. Chatbots can employ dialectical strategies to provide empathic responses, hear another person’s point of view and accept them without judging, and to allow people to express their feelings and concerns.

While chatbots have the potential to create individualized learning experiences and support performance, there are some factors to consider in using chatbots for learning.

• Manage expectations: Due to the limitation of natural language processing and machine learning algorithm, chatbot interactions could sometimes lead to awkward answers, frustrate users, and cause people to develop a general sense of mistrust. Keep in mind that even “intelligent” chatbots are limited to certain knowledge domains, and it is extremely difficult for chatbots to understand context, infer meanings, unpack ambiguity, and to have a useful memory.
• Focus on human-centered inclusive design: Chatbots’ replies to questions are based on their repertoire of previously encountered questions and answers. If most of the past questions came from the demographics of a certain majority (e.g., western culture, socio-economic class, etc.), chatbots will inevitably exhibit a bias toward the demographic majority in the responses and potentially underserve others. Humans need to work alongside the system to take ethics into account and to mitigate bias. Active monitoring and correcting these assumptions are urgently needed and require constant effort.
• Chatbots cannot replace humans: Humans converse in a way that chatbots cannot. We incorporate emotional intelligence in our decision-making process, we understand context to a larger extent, and make connections or draw inferences. While chatbots can engage and provide services at scale, freeing up time for staff to focus on more in-depth and complex interventions with learners, there are limitations in what the platform can do. Chatbots should not be a replacement for actual human communication. Rather, we ought to think of them as cooperatives, working side-by-side with us to enhance the overall workplace learning experience.