Chatbots are getting better and, more importantly, the tools that are needed to build chatbots and conversational agents are getting easier for non-developers to use. The first text-based conversational chatbot was ELIZA, developed as an MIT research project by Joseph Weizenbaum in 1966. It was more of a conceptual "toy" than actual profound research into systems that could hold real conversations with humans.
Since then, the use of text and voice-based assistants that can interact with humans in natural language has proliferated. According to Business Insider, the market for chatbots had grown to over $2.9 billion by 2019. Historically, the vast majority of these chatbots have been simple in nature and could not be classified as intelligent. But as developers incorporate AI, chatbots are advancing towards greater usability and more advanced capabilities.
Breaking down typical chatbots
Chatbots are built to provide application functionality that can interact with humans in a conversational manner through either text or voice format. Chatbots and conversational assistants are best used when users prefer conversational interaction over typing, swiping or clicking. This includes situations where back-and-forth interaction is more efficient as well as when physical input is inconvenient or impossible, such as while driving.
Since ELIZA, many more chatbots have been created. They typically follow a pattern in which they detect keywords and use prepared responses that can be combined in different ways for different circumstances. These bots also use connections to data to use previous answers for future responses.
However, these systems are, for the most part, not particularly intelligent, as real intelligence is more than detecting keywords and employing predefined sentences as answers.
Making chatbots more intelligent
AI is adding power to these chatbots and helping bridge the gap between humans and machines by employing natural language capabilities. These more intelligent chatbots are more capable and are being used in a variety of different contexts such as online customer support, phone interactions, information retrieval, assisting with online commerce or tech support, and the increasing popularity of voice assistants.
Because chatbots are easy to deploy, companies find them a great first use case for AI within their organization. Because bots can provide consistent results without the need to sleep or take breaks, companies are able to keep them deployed to engage and interact with customers. Organizations including banking, finance, retail, and others have AI-enabled chatbots to help enhance customer engagement, collect basic customer information and answer general company questions.
Many governments are seeing the benefit of these chatbots. The United States Postal Service has been using chatbots to help customers with a variety of tasks and the U.S. Citizenship & Immigration Services deploys an AI-powered chatbot Emma to help users answer a large variety of questions about various services, processes and forms.
The conversational interface has positively impacted countries and companies to make more efficient use of human team members and generate efficiency. Companies can drastically minimize the time it takes to resolve a query with the use of conversational assistants. These assistants enhance employee productivity by automatically following up on scheduled tasks. They also enhance the ability to handle and tackle client queries by networking with other chatbots.
Advancement of chatbot and conversational system tools and platforms
It is only recently that companies have been tapping into the power of AI and machine learning to build more advanced conversational applications. Today, organizations can find many offerings that provide templates and design options that make it possible for them to easily deploy chatbots to websites, mobile apps, voice assistants, messaging apps, or text interactions and interact across those channels simultaneously.
One of the earlier entrants into this market, Dialogflow (formerly known as API.AI and now part of Google), built a simple platform for developers as well as casual users to tap into existing NLP technology to build conversational agents. These could then be deployed to a wide range of endpoints from website chatbots to Slack and Facebook messengers, to voice assistants such as Amazon Alexa and Google Home.
Not to be outdone by Google, Microsoft released its Bot Framework which provides a platform to build and connect conversational bots interfacing with text on web and SMS, Skype, Slack, Office 365 mail and many other services, which can then be hosted in the Azure cloud environment.
Amazon also has chatbot building capabilities with its Lex offering, using the same technology that powers its Alexa voice assistant, and integrates with mobile devices, web apps, Facebook Messenger, Slack, and SMS through Twilio. IBM also offers rapid chatbot development through its Watson platform, and SAP acquired Recast.ai in 2018 to provide similar features.
Many other companies are in the market as well, focused on bot development. These include Chatfuel, Botkit, Botstar, Flow XO, Recast.ai and Motion.ai. Products such as those by Reply.ai keep the human in the loop by taking an "augmented intelligence" approach, feeding conversations that don't satisfy the human participants to other humans who can intervene. Other platforms offer a unified platform for creating bots with management tools and analytics dashboards that permit customers to build a general conversational service.
Chatbots have gone from being able to simply answer basic questions to offering truly beneficial support for users of all types. With the cost and complexity of chatbot development being significantly reduced, it's now a relatively trivial matter to implement conversational technology. With more widespread conversational systems, chatbots are ready to offer their assistance.