What is a chatbot?
A chatbot is a software or computer program that is designed to simulate conversation with human users. They are used extensively today, especially in fields like customer service where users ask similar questions over and over.
The very first chatbot, Eliza, was written by MIT computer scientist Joseph Weizenbaum in 1964-1966. The basic technology behind Eliza was to recognize key words or phrases and to respond to those phrases with pre-programmed responses.
Many chatbots today actually still operate using similar logic, but often these simple, rules-based dialogues aren’t sophisticated enough to handle many customer service requests. If the human doesn’t use the exact word the chatbot has been programmed to recognize, the chatbot won’t be able to understand.
Let’s say you ask your favorite clothing store’s bot if they carry a certain type of clothing. If you were to type, “maroon jumper” instead of “red sweater” and the bot wasn't programmed to recognize these synonyms, you wouldn’t find what you’re looking for. This problem has led to the advent of AI chatbots.
What’s the difference between rules-based chatbots and AI chatbots?
Rules-based chatbots are simple and have limited capabilities. They are programmed to recognize a keyword or phrase and deliver a canned response based on that input.
To avoid the frustration of the chatbot not understanding user input — like in the red sweater/maroon jumper example above — many rules-based chatbots are “button-based” and guide users down a decision tree where they are able to choose between a limited amount of answers.
AI-powered, NLP chatbots are far more sophisticated in their uses. AI chatbots use natural language processing (NLP) to determine the intent behind a user’s question. Instead of relying on keywords or buttons, users can type as they would talk to a human agent and the bot can understand the context and respond accordingly.
AI chatbots can often resolve requests without human interaction and they learn and grow as time goes on. This is why they are often referred to as virtual agents or intelligent virtual assistants, because they can respond in a human-like way and can resolve certain requests, especially simple or repetitive ones.
What's more is that in the last year, we've seen the capabilities of automation grow by leaps and bounds thanks to the dawn of generative AI -- the tech behind tools like ChatGPT. This has only served to further eclipse rules-based chatbots in their potential use cases within the customer support sector and beyond. Not to mention, with chatbots that use generative AI, you can get started with automation in just a few minutes. Simply connect to your knowledge source, and you're ready to go.
Interested in trying out a gen AI chatbot for yourself? Check out UltimateGPT, which harnesses the power of ChatGPT to bring you the best in CX automation.
Read more on the power of AI
What type of chatbot is right for you?
Rules-based chatbots can still be useful for task-specific situations. If you need a bot to handle reservations at a restaurant, for example, the bot is always asking the same few questions, with similar input from the user each time.
In this instance, the bot asks what time they would like to make a reservation for (perhaps with a limited amount of time slots available to choose from via buttons), how many people will be dining, and for a name and phone number. The task is then complete and the chatbot does not need to recognize any extraneous information.
The benefit to this type of chatbot is that they are relatively simple to build and cheaper to maintain. The downside is that they are unable to answer complex questions and they cannot be trained or improved over time.
AI chatbots and virtual agents are best for handling more open-ended questions. They are ideal for customer support teams who receive high volumes of requests, especially if some of them are repetitive queries. An AI chatbot can instantly resolve the simple requests or gather contextual information from the customer for an agent to reference later. They can also enable customer support teams to provide service in multiple languages or be available 24/7. And in the era of generative AI, they can do all of this in an uncannily natural, conversational way -- tailored to respond to your customer's mood and tone of voice.
The benefits of AI chatbots are that they are able to understand human language without explicit programming and they can improve and learn as time goes on. The downside to this type of chatbot is that they require more time and resources to get up and running than a simple rules-based bot.
Examples of chatbots
A rules-based chatbot example
One fantastic example of a rules-based chatbot, is the Domino’s Pizza Chatbot. Because Dominos customers are always accomplishing the same simple task (ordering a pizza), a rules-based chatbot is ideal. But Domino’s took it a step further and set up a rule that made ordering pizzas easier than ever.
If a user who had their “favorite order” saved on their profile sent an SMS of the pizza emoji to the Domino's Pizza number, that order would be placed immediately with no further action needed from the customer.
An AI chatbot example
A great example of an AI-powered chatbot is the Finnair virtual agent. Finnair has used a virtual agent since 2018 to streamline their customer service and provide support in multiple languages, including Finnish and English. So when the Covid-19 pandemic hit, they were able to handle it with relative ease.
Building on existing chat automation processes, Finnair simply tweaked their AI chatbot’s welcome message to provide Covid-19 information right away. Customers could indicate whether their query concerned Covid-19 or not, and were able to go down an automated support path regarding refunds and cancellations without needing to speak with a human. In the first few weeks, the automation rate reached 50% due to the increase in volumes directly related to Covid-19.
Whether your team ends up going with a simple chatbot or an AI chatbot, both will allow your customers to self-serve to a certain extent and free up your customer support team to spend time on more meaningful tasks.