A chatbot is an artificial intelligence (AI) program that creates conversational interaction between the chatbot and another user through voice commands or text chats.
A chatbot is also known as an Artificial Conversational Entity (ACE), chat robot, talk bot, chatterbot, or chatterbox. A user can ask a chatbot a question (What's the weather today?) or make a command (Play Rick Astley), and the chatbot responds or performs the action.
Early versions of chatbots included the iconic ELIZA program (1966) which simulated a psychotherapist, and PARRY (1972), meant to mimic a person with paranoid schizophrenia. Both of these chatbot programs simulated typed conversations.
Now, many chatbots include voice interaction that leverage natural language processing (NLP) and Natural Language Understanding (NLU. This gives them complex functional features, like the ability to carry out a task or gather information for a user.
Modern chatbots are often used where simple interactions and limited responses are required. They play a significant role in customer service and marketing applications. A chatbot can answer questions about products, services, or policies. If a customer needs to do more than the chatbot can handle, the program can escalate to hand off the interaction to a human operator.
Modern chatbots are also often used online and in messaging apps. They are also included in operating systems as virtual intelligence assistants (e.g., Apple's Siri and Windows' Cortana). Chatbot appliances like Alexa and Google Home are also growing in prevalence and can perform a myriad of functions based on user commands.
Chatbots may function in two main ways, depending on if they are built on a set of rules or machine learning.
A chatbot based on a set of rules is quite limited. It can only respond to a limited set of precise commands. If a chatbot is programmed to tell you the time or the weather, it will not be able to respond to a request like "Which artist wrote 'Never Gonna Give You Up?'" These chatbots are often stateless, meaning that they approach each interaction as if it were with a new user. These bots have no active memory of prior communication.
A chatbot that functions using machine learning is built on artificial intelligence (AI) technologies, including natural language processing, deep learning, and machine learning algorithms. These complex chatbots understand language, not just specific commands.
AI-powered chatbots require a massive amount of data and are stateful, meaning that they can review past interactions and include that context in its responses. The more a stateful bot interacts with its users, the better the bot gets at recognizing human speech and predicting the appropriate response or action to take.
Some examples of chatbots you might interact with on a daily basis are virtual assistants like Google Assistant & Amazon's Alexa. Messaging apps like Facebook messenger or WhatsApp also often leverage chatbots for businesses to better communicate with their customers.
While bots appear vastly different, they fall into two main categories: transactional (stateless) or conversational (stateful).
- Transactional or Stateless bots don’t require history – every request is treated as discrete.
- Conversational or Stateful bots rely on history and information collection to complete tasks.
The time savings and efficiency gained from AI chatbots and conversational assistants help companies increase their customer service productivity or sales. By leveraging chatbots, human agents are freed up to focus on high-profile customer service or sales rather than spending all day answering questions about a business' return policy.
Chatbots also help improve the customer experience. Many businesses have leveraged phone trees for decades, which work but are time-consuming for the customer to navigate. A chatbot can turn "Press 1 for Alex, press 2 for Joey" into "Who can I connect you with?" The chatbot can understand the caller's input if they say "Joey" and route the call directly.
Today, people are using messenger apps more than they use telephones or social networks. Providing chatbots on these messaging platforms opens up large markets to a business.
Interested in learning more about chatbots, or even building one yourself? Read our complete guide to intelligent bots for all the chatbot resources you need in one spot.
Once you're ready to build, the following resources may help you out:
- Read how chatbots can help you scale your customer communications
- Leverage Twilio Studio to build a chatbot that can take a coffee order
- See how you can build an SMS-powered bot with Node.js
- This Udemy course teaches you how to build chatbots from scratch