Demystifying Chatbots and Virtual Assistants
Know your robot: the difference between virtual assistants and chatbots.
Meet Sancho. He’s the still-slightly-secret thing we’ve been working on here at LigaData. When we first began crafting his identity, he looked like this little guy:
Clearly, very robotic. But the thing is, he talks (or chats), and before you know it, he’s developing a persona. Sometimes it’s remarkably like our CEO’s. Other times, we aren’t quite sure who’s personality he is taking on. Now he’s looking more like this (he has a hat with a feather, and is riding a mule, in case you were wondering).
But what is he even? A chatbot, a virtual assistant, or something more?
In this post, we’d hoped to shed some light on the difference between chatbots and virtual assistants. Everyone has a different perspective, and there are few agreed definitions. What is clear is that a very basic chatbot sits at one end of a continuum, and gets more complex in functionality, language and technology as chatbot tools morph into virtual assistants. They are both apart of the humanoid companion interface family which unites them and humanoid robots under one umbrella according to this piece by neurolab.
Some may use the phrase ‘conversational agents’ as a description both of chatbots and virtual assistants; others (TechTarget) identify conversational agents as discrete applications that do more than chatbots but less than virtual assistants.
Rob High of IBM says:
“A conversational agent is more focused on what it takes in order to maintain a conversation. With virtual agents or personal assistants, those terms tend to be more relevant in cases where you're trying to create this sense that the conversational agent you're dealing with has its own personality and is somehow uniquely associated with you.”
But the idea of a continuum of conversational agents works for us; a continuum that includes the following elements:
Scope Traditional chatbots focus on clear, restricted tasks (and cannot compute if asked to do something outside of what they’ve been programmed to do). Virtual assistants, while often designed to do specific functions (virtual customer assistants, virtual personal assistants, virtual employee assistants, etc) can better handle the unexpected; whether that’s in the user phraseology or topic. They can initiate actions, based on their learning, and sometimes they dont need to involve the user at all.
Memory Chatbots consider each conversation a brand new conversation, and have no ability to take into account an earlier conversation, or knowledge about its interlocutor or correspondent. Virtual assistants can equip themselves of customer data, transactions and conversations to result in a more intelligent exchange and range of activity (such as finding contacts and scheduling meetings). Interacting with a chatbot can be like talking to a stranger every time, whereas a with a virtual assistant you know there will be some recollection of the conversation.
Technology With chatbots, users interact with a rule-based process, with little application of AI. But it’s the introduction of machine learning that brings potential to the chatbot/agent tool, enabling it to discern and predict the user intent and deliver benefits, hopefully, both to the business and its users. This delivery of machine learning makes the chatbot transcend into a virtual assistant.
“Gartner: VAs use semantic and deep learning (such as deep neural networks, natural language processing, prediction models, recommendations and personalization) to assist people or automate tasks. VAs listen to and observe behaviors, build and maintain data models, and predict and recommend actions.”
User experience It should not take long for a user to understand that with a chatbot, they are engaging with an automated system – and indeed, it’s important to manage user expectations. The more sophisticated an agent becomes, both in language and comprehension, the greater the risk that a user will believe it’s engaging with a human, which can open a whole can of worms.
The features they share are include the use of Natural Language Processing (to a greater or lesser extent), of being available 24/7, of taking inputs both by text and voice (one article describes “text-processing chatbots and speech-processing virtual assistants”, but we don’t believe that distinction is important). In fact, they are becoming increasingly multi faceted – taking into account facial expressions, movement, and tone of voice to enhance understanding of the user’s intent.
For all the amazing examples of conversational agents at work today, there is still so much evolution to come. Gartner is particularly scathing about the present state of affairs, at least in the world of customer service:
“The current generation of VCA (Virtual Customer Assistant) deployments and other types of conversational agents are often implemented incorrectly, failing to capture customer intent or handle unexpected input elegantly. Only VCAs that create a compelling user experience and deliver business value will survive.”
And surely that can be said of everything we do.
And as for Sancho, what is he? He likes to consider himself a virtual assistant. Sophisticated enough to understand what you are telling him and insightful enough to learn terms he doesn't understand.
When you meet him, you decide… To register for our BIO (By Invitation Only) group to be able to give Sancho Analytics a test drive for yourself, click here to register your interest, or visit our homepage.