[cs_content][cs_element_section _id=”1″][cs_element_row _id=”2″][cs_element_column _id=”3″][cs_text]
A chatbot is a computer program, powered by rules, and sometimes artificial intelligence, that is designed to carry on conversations with humans. Chatbots are one of the digital technologies Rehabit Labs are using as part of our repertoire of behavioural influence tools. We think that the future of digital interfaces will involve a combination of the traditional visual user interface (UI) and conversational interfaces.
There has been a massive proliferation of chatbot technology in the last year or so. This growing popularity is the result of a number of different causes including the growing popularity of messaging platforms (eg. Facebook Messenger) and the growing power of AI and machine learning; bots are the new apps.
The sophistication of chatbots exists along a broad spectrum. If you are thinking of incorporating a chatbot into your website, social media presence or digital product there are 2 different questions that are useful to consider. The first is the memory of the chatbot; how much it remembers from one interaction to the next. The second is the capability of the chatbot; how effective it is at understanding and providing a solution to the explicit or implicit intent of the person interacting with it.
Memory is something we take for granted in conversation. When I ask you “do you have children?” and you say “yes”, and then I ask you “how old are they?”, memory is implicit. This is because the second question assumes that we both remembered the first question and its answer. Not all chatbots have memory (technically this is often referred to as “state”).
Chatbots range in memory. At one end of the spectrum are chatbots that have no memory at all. They don’t remember anything between one interaction/response and the next. Totally memoryless chatbots process each message as if it is coming from a person they had never seen before.
The next type of chatbot is one that remembers what you just said. A good example of this type of memory are automated phone menu systems. The choices you make at each menu are “remembered” and then another sub-menu is offered. These chatbots limit their memory to a single conversational session with the person. They provide better utility than chatbots with no memory at all, however the experience still feels rather machine-like.
The most powerful chatbots are those that remember you when you come back. These chatbots keep a memory of each conversation you have had. This enables them to recall what you asked for before, and adjust their response based on that. The more that you interact with the chatbot the smarter it can potentially become. This provides the opportunity to provide a more personalised experience and to better mimic the quality of a real conversation.
You can see from these examples how memory is a critical attribute of a chatbot and the user’s experience. The other attribute that is critical to a chatbot is its underlying capability.
[/cs_text][x_image type=”none” src=”https://www.rehabitlabs.com/wp-content/uploads/shutterstock_524733751-copy.jpg” alt=”” link=”false” href=”#” title=”” target=”” info=”none” info_place=”top” info_trigger=”hover” info_content=”” style=”height:400px;display: block;margin: 0 auto;”][cs_text]
The purpose of a chatbot is to address the user’s intent in interacting with it. Some intents are simple and deterministic – eg. “what’s the weather forecast like for today?” or “is fasting a good way to lose weight?”. This class of capability is one of the most common uses of chatbots in which simple question/answer pairs, like those found in a FAQ (Frequently Asked Questions), are “botified”. This class of chatbot is generally powered by simple rules that match keywords in the user’s message with matching content or data. Rule-based chatbots are relatively cheap and quick to implement and work well in situations where requirements are relatively basic.
Other intents are more complex and nuanced – eg. “can you offer me advice on improving my nutrition and physical activity habits?” or “I’ve just been diagnosed with diabetes. What should I do?”. The more complex the domain of the intent, the more sophistication the chatbot needs in order to understand the user’s intent and a provide helpful response. This is especially so when the domain involves human behaviour. Improving someone’s knowledge alone has been shown to be limited in terms of its ability to effect behavioural outcomes. Recent international findings strongly support an approach that supports the person’s overall behavioural capability. This requires a chatbot that has a more sophisticated understanding of the user, and how to effectively bring an understanding of the problem domain to address their needs.
This class of chatbot is powered by AI (specifically machine learning) and NLP (Natural Language Processing). AI-based chatbots can not only gather information but also understand the intent of questions and their context. AI-based chatbots also have the ability to learn as they go rather than simply responding to a prescribed set of questions. This makes them a better long term solution for fulfilling multiple functions.
The power of conversation
At Rehabit Labs we see chatbot technology as a tool that should be treated as an integrated element of a larger digital technology toolset. We believe it is important to recognise where the strengths of chatbots lie, and also where they don’t. This enables chatbots to be positioned within a coordinated strategy that contributes towards strengthening the organisation’s overall capability in shaping behaviour.
Chatbots for instance provide a very accessible way for users to get information. A single conversation however is rarely powerful enough to change behaviour. In order to change behaviour you need multiple conversations. This calls for a strategy based on building a digital relationship and an ongoing engagement program that uses AI to adapt and personalise to the user. Chatbots here can play both a tactical role of meeting the user’s immediate informational needs, as well as a strategic role in establishing a relationship with the user in order for the organisation to be a more effective agent of positive behaviour change.