The ultimate glossary of conversational AI terms

Conversational AI is a rapidly growing market that is powering automation for a variety of industries and using a wide range of technologies. A recent report estimates that the conversational AI market could exceed US$30 billion globally by 2030.

With so much happening in this exciting field, it’s sometimes hard to keep up with the wide variety of ways people are talking about conversational AI. It can get confusing.

Brush up your knowledge of the growing bank of terminology used to talk about conversational AI with our handy glossary of terms.

artificial intelligence

Commonly abbreviated to AI. This is a general term that refers to a broad set of technologies that allows software and machines to perform ‘human-like’ reasoning and interactions. One of the defining characteristics of AI. is its ability to convince humans that they are interacting with a human-like thinking entity. Another defining characteristic of AI. is its ability to ‘learn’ independently from its initial programming and adapt to new environments.

Related terms

machine learning

AI assistant

AI assistants are software-based services that use some form of artificial intelligence to help automate processes or perform tasks at the request of human users. AI. assistants are a type of virtual assistant. Often an AI assistant will take the form of a voicebot or chatbot to facilitate interaction with human users.

Related terms

virtual assistants, voicebot, chatbot

automatic speech recognition

Automatic speech recognition (ASR) is a software-based service that helps convert pure phonetic input from a human speaker into recognizable words. Speech recognition is often used in conjunction with natural language processing, although speech recognition may be used outside of natural language processing services to power, for example, interactive voice response services where simple voice inputs from humans can be used to navigate a menu system via a telephone.

Related terms

natural language processing, interactive voice response, speech-to-text conversion, voice recognition

chatbot

A chatbot is a software-based service that enables humans to ‘talk’ with a service to either retrieve information or perform a transaction. Typically users interact with the chatbot by typing questions into text interfaces such as an SMS or social media chats or a chatbot on a landing page. A chatbot will use natural language processing to determine what help or information the user needs. Increasingly, chatbots are being distinguished from voicebots although sometimes the two terms are conflated or used interchangeably. Chatbots enable business process automation and are frequently used in contact centers or on social media channels. For more about chatbots, see Toptal’s guide to chatbot terminology.

Related terms

voicebots, natural language processing

conversation design

Conversation design is carried out by a conversation designer when creating a virtual assistant based on voicebot technology. The designer creates a structured conversation for the voicebot to follow to facilitate business automation. The designer also sets up criteria for the voicebot to smoothly handle interruptions and diversions from the scripted conversation by human users. Voicebot AI technology allows the voicebot to maintain a continuous flow of conversation while dealing with various levels of ambiguity that cannot be handled by older forms of automation.

Related terms

voicebots, natural language processing, intent

conversational AI

Conversational AI is a subset of technologies for artificial intelligence. It refers to types of AI software that are designed to handle voice-based interactions with humans, usually using natural language processing and speech recognition and synthesis combined with other automated services. Voicebots typically make use of conversational AI. Siri and Alexa are two widely used examples of voicebots that use conversational AI.

Related terms

natural language processing, speech recognition

conversational IVR

Conversation IVR is a type of conversational AI that uses speech recognition to guide a human user through a branching menu system of information. Instead of requiring users to key a number from their phone dialler interface, conversational IVR allows users to vocalize their selection or response. Whereas traditional IVR systems had speech recognition technology to handle simple voice commands such as “yes” or “no,” conversational IVR allows people to communicate their inquiries in more complete phrases via a natural language understanding

Related terms

conversational AI, speech recognition

customer experience automation

Customer experience automation refers to a set of strategies that a company or business uses to automate interactions with customers. Customer experiences can and frequently do include customer service transactions but also embraces a wider set of considerations for how the brand communicates with customers. Automation can, for example, be used to send messages to customers according to a communications plan or reward customers automatically based on customer behavior records stored in the company’s customer relationship management database (CRM).

Related terms

customer service automation

customer service automation

Customer service automation is a subset of customer experience management. Contact centers will make use of customer service automation to automate certain high frequency, predictable customer interactions to improve access to the service, optimize efficiency or scale customer service levels instead of outsourcing services to a BPO organization.

Related terms

customer service automation

entity

An entity is a key piece of information given by a user when interacting with conversational AI. Named Entity Recognition (NER) along with intent recognition are two of the most important systems used by virtual assistants. It allows a virtual assistant to recognize and capture key information such as times, dates and names that are relevant to completing a transaction or request. For example, if a customer is booking a cab, the virtual assistant will capture the time and location for the cab pickup using NER.

Related terms

intent

intelligent automation

Intelligent automation is a typically software-based service that automates tasks or processes, usually using some form of artificial intelligence. The software will display some form of decision-making capability and in sophisticated applications, the ability to learn from unexpected inputs or errors. Intelligent automation can also refer to types of interactive software that give the appearance of having human-like intelligence, such as is the case with voicebots, chatbots and some types of virtual assistants based on conversational AI technology.

Related terms

customer service automation, conversational AI

intent

In conversational AI, intent is used to refer to the intent of a human user when communicating with the AI. In other words intent is a specific action the user wants the bot to perform. Users can formulate their requests in various forms, but those requests often have the same intention. Conversational AI systems use natural language processing to disambiguate users’ utterances into specific intentions. A conversation designer designs the voicebot in such a way to help the AI recognize human intent by training the AI to notice certain patterns such as key phrases and link them to appropriate responses for the AI to give. For example, if a customer is communicating with a voicebot and interjects that they are quite busy now, the AI will recognize the intent (that the human wants to postpone the interaction) and can simply reschedule the conversation for another date or time.

Related terms

conversational AI, conversation designer, natural language processing, speech recognition

interactive voice response

Interactive voice response (IVR) is a type of software-based service that uses speech recognition or natural language processing to enable humans to interact with a voice-based application or service. A common application of IVR is to allow a human caller to navigate an audio-based menu with simple vocalized statements rather than pressing keypad numbers on their telephone dialler.

Related terms

speech recognition, natural language processing

machine learning

Machine learning is a set of techniques and technologies that help software-based applications make decisions, improve performance or recognize patterns with increasing accuracy without a high degree of human intervention or feedback. Most machine learning requires two sets of data: training data and testing data. The algorithm is trained using the smaller training data set and then tested using the testing data set to assess the accuracy with which the algorithm has been trained. Generally speaking, the greater the accuracy required of the algorithm, the larger the data sets required. While machine learning can take many forms, a commonly encountered way of facilitating machine learning is with a neural network.

Related terms

artificial intelligence, neural network

natural language processing

Natural language processing (NLP) is a  set of machine learning methods that allow human speech to be understood in relatively unstructured contexts by a machine or software. NLP is often used as a type of voice user interface between humans and a software service that allows the human to speak naturally and freely to direct the software to provide information or perform tasks. NLP may need to be trained to recognize the speech patterns of the humans it interacts with. NLP is often used  in pair with speech recognition although NLP can entail more complex handling of language inputs beyond recognizing the words spoken.

Related terms

machine learning, speech recognition

Source: Toptal’s guide to chatbot terminology

neural network

Neural networks are a set of technologies that are commonly used in machine learning and predictive modeling services. Neural networks are characteristically designed to be similar to the synaptic structures found in brains. Neural networks are capable of types of non-linear analysis that other types of decision-making models followed by machines struggle with, such as image recognition and speech recognition where the inputs may be entirely novel to the algorithm.

Related terms

machine learning

speech-to-text conversion

Speech-to-text conversion (STT) is a software-based service that uses speech recognition technology to convert human speech into text. Speech-to-text conversion is often used to produce transcripts of conversations between humans and voicebots. These transcripts can then be stored and analyzed. In a virtual contact center, for example, speech-to-text conversion can be used to produce text-based records of customer queries that can be collated and analyzed to identify trends related to customer needs or satisfaction with the service.

Related terms

speech recognition, natural language processing, interactive voice response

virtual agent

A virtual agent is a type of software-based automation often used by contact centers to handle common customer inquiries or transactions. Tasks commonly performed by a virtual agent include routing a call to a human operator specialist, tracking an order status, making or changing a booking or appointment for a customer. The term virtual agent is sometimes used interchangeably with virtual assistant. Voicebots are often used to create virtual agents that can help automate a wide range of services offered by a contact center.

Related terms

virtual assistant, virtual contact center, voicebots

virtual assistant

A virtual assistant is a software-based service that helps humans perform tasks or access service information without the need for human assistance. The term is sometimes used interchangeably with virtual agent, although virtual assistants can be found in a wider range of contexts. Siri and Alexa are commonly encountered examples of virtual assistants.

Related terms

virtual agent, virtual contact center, voicebots

virtual contact center

A virtual contact describes a contact center that relies partially or entirely on software-based services to resolve customer queries and issues, such as virtual agents. A virtual contact center is typically composed of a mix of virtual agents with virtual agents handing over more complex or demanding customer queries to specialist human customer service agents. This allows the contact center to handle higher volumes of customer queries without significantly increasing costs and without compromising on the quality of customer service.

Related terms

virtual agent

voice recognition

Voice recognition is a software-based service that recognizes when a human is speaking with the service. It is a type of speech recognition. Voice recognition can be used solely to identify the voice print of particular users whereas speech recognition is focused on recognizing the words being spoken.

Related terms

virtual agent, virtual contact center, voicebots

voicebot

A voicebot is a software-based service that uses speech recognition and natural language processing to enable a human to interact with the service to perform tasks or retrieve information. Voicebots have a wide range of applications such as being used as the underlying technology for virtual assistants or virtual agents.

Related terms

virtual agent, virtual contact center, virtual assistant, speech recognition, natural language processing

voice user interface

A voice user interface (VUI) is a mode of interaction that relies on speech recognition of natural language processing software to interact with a computer-based service by speaking rather than typing. A VUI is contrasted with a GUI (graphical user interface) and not to be confused with a visual user interface which is a type of graphical user interface.

Related terms

speech recognition, natural language processing

If you would like to learn even more about conversational AI and how it could help grow your business, feel free to get in touch with the Talkie.ai team or book a demo. We’re always happy to talk!