Chatbots vs. Conversational AI

February 22, 2021 • 7 minute read

For real business value, companies should carefully consider whether scripted chatbots are too limiting to achieve their goals, versus the personalization power of Conversational AI agents.

What is a Chatbot?

Chatbots are AI-based programs that facilitate basic conversations between humans and computers. According to independent research firm Ovum, chatbots are “best deployed where there is a clearly defined scenario that is predictable and capable of being scripted in advance, with a clearly structured or predictable flow to the conversation; for example, answering frequently asked questions (FAQs) or engaging in chat interactions to process payments.”

Think of a customer-facing business process or request that a company handles multiple times each hour every day. Locating store hours, product descriptions, inventory and more have become as easy as typing in a request. Still, companies spend millions of dollars every year fielding these kinds of queries, most often with humans driving these interactions. Call center workers, customer service email respondents and even in-store associates provide the same responses over and over again: “Yes, we have this shirt in a size large,” or “Yes, we’re open on Presidents’ Day.” These basic questions cost a lot of money to answer at scale with live, human agents.

While customers jam up phone lines and inboxes with easily answerable questions, other customers wait on hold. Although quantifying lost revenue from inordinately long hold times may vary from company and industry, it’s obvious that dissatisfied customers, and customers who drop off calls out of frustration, translate into lost business. Chatbots can handle thousands of questions at a time and respond to most of them instantly. Those they can’t answer are escalated to human workers who would otherwise have been busy responding to simple FAQs.

What Are the Differences Between AI Chatbots and Conversational AI?

Companies that deploy plug-and-play chatbots soon realize that chatbots suffer from severe limitations. Because chatbots use basic Q&A scripts to serve users, they’re incapable of performing complex tasks and solving unique problems. This inevitably leads to elevating chatbot interactions to already-overwhelmed human service agents.

Even if one looked beyond the inherent flaws associated with a chatbot’s simple dialogue scripting, it’s clear that a perfectly scripted chatbot experience cannot be applied from one brand to another without training, customization and true intelligence. A company’s business processes, its product catalog, its service agreements are all unique and complex — and that’s before you add user personalization into the equation.

Why Are Chatbots Basically Just Talking Bots?

Conversational AI has been designed to understand user requests via natural language processing (NLP); it can absorb information, learn by following process maps created from prior interactions, observe colleagues to discover the optimal course of action, and apply supervised learning to address similar scenarios without human intervention. These capabilities are significant in performing tasks beyond the most basic customer service engagements.

The entry point into a dialogue with Conversational AI is through what is called a user utterance. That user utterance can be delivered via a chat box on a website, a voice call either through a mobile or a home assistant such as Amazon Echo, within a chat app such as Facebook Messenger, or wherever a company would like customers to interact with AI. Before Conversational AI ever speaks to a customer, Business Process Modeling and Notation tools model a company’s processes to ensure that Conversational AI does exactly what the company wants (e.g. Send a replacement card to a customer who lost his card). But what happens when someone enters an interaction with multiple intentions?

One key difference between Conversational AI and chatbots is in complex intent recognition capabilities. Conversational AI uses neural network algorithms to detect intent. If a banking customer says, “I lost my credit card yesterday,” Conversational AI will remember its training as a credit card replacement agent. It will know that in the case of a lost credit card, the customer’s intent is typically to deactivate the missing card, get a new card issued and resolve any disputed charges.

For a typical chatbot that isn’t armed with cognitive intelligence, multiple intentions cause confusion. Conversational AI can not only register multiple intentions; it can also triage them to ensure the most important processes are handled first.

The Science Behind Conversational AI

The “recipe” for building Conversational AI is a mix of complex and proprietary technology. True Conversational AI systems use natural language processing and understanding (NLU) as the core of the platform. Working in conjunction with NLP and NLU is a proprietary blend of multiple Deep Neural Networks (DNNs) and natural language data sources that provides the system with the ability to contextually understand and interpret simple and complex multi-sentence requests.

NLG-based Clarification enables Conversational AI to dynamically drive open conversations to identify what the customer would like the system to resolve for them. This includes asking clarifying questions or asking them to elaborate specifically on a topic. Most importantly, this dialog is not programmed but generated dynamically based on the actual dialog with the customer. Chatbots are not capable of conducting conversations that are not programmed.

Working memory lets Conversational AI connect relationships to vast repositories of metadata, which enables the system to provide personalized support over the course of time. Chatbots treat every interaction with a user as if it were the first interaction. Episodic memory provides cognitive access to previous conversations and allows systems to assist human colleagues with recommendations, based on the collective experience of human agents.

Conversational AI stores facts, concepts, and the associations between them in its semantic memory. From standard operating procedures to policy documents, it can be trained to apply them to conversations. It also uses state-of-the-art Affective Computing and Sentiment Analysis techniques to continuously model user's emotion, mood and personality. Chatbots offer no semblance of human emotion and often deliver “robotic” responses.

Conversational AI can dynamically navigate business process flows, without having to follow a step-by-step process, to achieve a desired outcome. This allows it to jump from one subject to another during conversations with users. Symbolic logic allows Conversational AI to dynamically extract variables from utterances in user conversations, and use those as part of its interactions for clarity and engagement.

Conversational AI Chat Leads to Complex Automation

Conversational AI also understands the ins and outs of the business for which it is deployed. For example, banking Conversational AI is an expert in investing, credit card management, bill payments and more. Conversational AI understands banking regulations, common banking language, and it has a complex understanding of what customers require when they interact when a banking customer service rep. That’s because Conversational AI can integrate with data systems and business process models to learn and master specific products, processes and customer service requirements. It learns what to say and how to help customers transact, and it learns where to find information, both personalized and brand-specific, in order to provide customers with correct answers. If this information changes, the system will learn about the change and instantly apply it to interactions.

Low-level chatbots react to simple keywords or specific phrasing, while advanced Conversational AI systems are capable of discerning user intent from a wide spectrum of human utterances. This flexibility is particularly important when engaging with a large and varied customer base which unsurprisingly will have multiple ways of communicating the same idea: “How many minutes does the Premium Gold plan offer?” versus “How long can I talk on your top-level plan before you charge me extra?” versus “Does that Premium Gold plan give me at least two hours of free talk per month?”

Conversational AI Assistants and Hybrid Workforces

Conversational AI functions as a de facto translator of a company’s enterprise systems. Rather than requiring employees to remember the URLs, user names, passcodes and interfaces of dozens of enterprise systems, Conversational AI can connect to these systems to help workers enter and find information.

An employee who wants to log vacation days no longer needs to locate a hard-to-find or infrequently used human resources (HR) tool, remember their user name, reset their forgotten password, and attempt to recall how to navigate to the vacation request form. The employee can simply speak their request to the system and it instantly logs vacation days, sends the request to the employee’s manager and logs the manager’s response.

Think of the time employees waste entering, exiting and navigating enterprise systems. An HR system is among the more basic systems people use at work. Navigating IT helpdesk, finance, CRM, ERP and security systems can be exponentially more complex, even for employees who have been at a company for years. By connecting intelligent virtual agents to back-end software, a company removes the barriers and latency between intent and action. As Conversational AI is system-agnostic, it can seamlessly integrate with any enterprise process, system or layer. Regardless of scope or scale, it can be tailored to handle every business process requirement.

Simple chatbots are effective at Q&A-style interactions, but they cannot autonomously enter a new client’s information into a CRM tool, and then, unprompted, send a follow-up email in three weeks – chatbots are simply not designed to perform such tasks.

Businesses using Conversational AI spend less money hiring staff to handle basic tasks, and rededicate that money to hiring and training employees to answer more important questions. Customers no longer wait on hold for long periods of time, receive immediate positive responses through a virtual agent, and in turn tell their social networks about their positive experiences.

A chatbot will probably be able to handle some of these questions, but will still require constant escalation to a human worker due to the technology’s inability to decipher idiomatic questions, quickly learn new products and processes, and handle multi-context requests. What’s the point of hiring digital labor if human labor is still required to resolve simple issues?

Customer Service Chatbots Are No Substitute for Hybrid Workforces

Conversational AI is no substitute for human ingenuity. With intelligent virtual agents automating mundane tasks, humans have more time to think, brainstorm, test and imagine. Shifting employees from repetitive to challenging tasks enables them to think up new business processes, products, services and more. These employees understand what works and what doesn’t with regard to their company’s business processes and product offerings. By shifting them from repetition to conception, a business can build a qualified team of research and development experts with thousands of hours of collective experience.

Conversational AI shoulders the repetitive work that needs to be done, and it does so at scale and at machine pace. By doing so, it frees humans to handle more creative and high-value work, such as critical thinking and imaginative problem-solving. Additionally, Conversational AI is able to learn from human workers by observing their actions. When a human performs a task that the system is incapable of performing on its own, it studies the human agents’ actions, recommends a new resolution and asks a human manager if it can begin to perform the actions itself.

With a team of AI-powered experts who know the ins and outs of their company, businesses can dedicate their time to fixing old processes and products, and creating newer and better ones. Better processes mean more efficiency, which means fewer wasted resources and higher profit margins. Better products mean more sales, more customers and more revenue.

Automating tasks does not truly enable human creativity unless humans are fully freed from rote tasks. If they’re required to constantly oversee and remediate work done by simple chatbots, a business will only realize minimal benefits. These are just some of the ways that Conversational AI can improve a business in better and more substantial ways than a chatbot. Collaboration between humans and machines is the ultimate opportunity for today’s enterprise, but the level of the machines’ intelligence will ultimately impact how much a business improves. Settle for a chatbot and a company will reap limited rewards. Employ Conversational AI and the opportunities for improvement are far greater.

Next

A Beginner’s Guide to Conversational AI

Crossing the bridge between digital assistants/chatbots and real Conversational AI requires a fuller understanding of how the technology works and its potential business value.

In our latest white paper, A Beginner's Guide to Conversational AI, we explore these subjects for companies pursuing a near- or long-term technology strategy that includes Conversational AI solutions and Digital Employees.

Download our white paper to learn how to generate business value with Conversational AI.

Download