Chatbots can’t deliver value at enterprise scale like a cognitive AI platform
While chatbots have been around in some form since the 1970s, they’ve become increasingly visible in recent years. Chatbots fall under the umbrella of Artificial Intelligence (AI), but in order to truly enhance the user experience (UX) and build intelligent workflows, enterprises need to tap the power of a robust cognitive agent like Amelia.
A cognitive agent (sometimes referred to as a “digital colleague”) is able to discern a wide spectrum of human communications, assist users in executing complex actions, and improve at tasks over time. Chatbots, on the other hand, are lower-level and far more limited in their scope. You can read some detailed posts about the differences between the two here, here and here.
In the meantime, here are four substantial ways that a digital colleague delivers more business value than a low-level chatbot.
1. Chatbots chat, digital colleagues have conversations
It is possible to have a completely comfortable and satisfactory interaction with a chatbot — so long as a user doesn’t veer outside of the bot’s programmed script. Chatbots follow static decision trees to answer pre-written responses to specific inputs. However, if the user asks the same question in a way that the chatbot wasn’t pre-programmed to understand, e.g. “Hey, what time ya got?” versus “What time is it?”, the chatbot may not know how to respond. This lack of flexibility is a barrier for use within an enterprise that must accommodate a large variety of users, inputs and conditions.
Cognitive agents have advanced natural language understanding (NLU), which allows them to process human communication in many formats — even ones they haven’t been specifically programmed for.
Cognitive agents have advanced natural language understanding (NLU), which allows them to process human communication in many formats — even ones they haven’t been specifically programmed for. For example, in the “What time ya got?” example from above, Amelia would be able discern that this was a question and notice the word “time” and determine the user’s meaning. She also comes with a level of “social talk,” which allows her to converse with users who talk informally.
“NLU is all about understanding what the user means, not just what they say,” said Allan Anderson, IPsoft’s Director of Enterprise Solutions. “This is where disambiguation, co-referencing, negation, etc. all play an important role so that the context of a statement can be derived accurately. Getting the context correct is paramount as it impacts how the virtual agent drives the dialog with the user more appropriately through dynamic natural language generation of clarifying questions, which are not scripted.”
2. Digital colleagues can navigate conversational chaos
Humans are complex beings and regularly switch from one topic to another. Even during structured interactions like placing an order for a product or service, they can jump between subjects and change answers. For example, a customer at a diner may have just finished giving a waitress their hamburger order, but then say, “You know, actually, can we make that a chicken sandwich instead?”
These jumps are handled with ease by most humans — it would be no problem for a waitress. However, this “context switching” is something a static chatbot would find difficult to handle. Imagine the frustration if the waitress told the customer, “I’m not sure what you mean, let me ask my manager.” A robust cognitive agent, on the other hand, is able to apply “actually, I’d rather” edits to earlier parts of a process, while keeping other information the same.
For example, a customer ordering a new grey laptop with upgraded RAM and a backlit keyboard could decide at the last moment on a different color. A digital colleague like Amelia would be able to process this one change from an earlier step without touching the rest of the customer’s order. This is a key differentiator from chatbots, and can provide an advantage for companies that use cognitive agents. The less “human-like” the interaction with a chatbot, the more frustrated a user will become, and the less likely they will use that service again.
3. Digital colleague, teach thyself
Not only do digital colleagues come with more functionality than a chatbot, they can improve at tasks over time through machine learning. That means every interaction they have with a human allows them to expand their knowledge base.
To continue the time example mentioned above, with experience Amelia might be able to understand that humans sometimes informally end questions with a verb (“Hey, what time ya got?”). In the event that Amelia is not able to assist a user, the session can be escalated to a live agent, while Amelia silently observes the transaction in the background and draws connections between the unrecognized input and the eventual satisfactory conclusion. A digital colleague has the ability to independently improve at their function over time, whereas a chatbot doesn’t improve without direct human mediation.
4. End-to-end functionality
One of the most important differentiators between a virtual agent and a chatbot is its ability to independently assist users from the beginning of a process to its resolution. Whereas chatbots often point users toward an existing webpage, video, or other information (in basic FAQ-answering functionality), a digital colleague can tap into its backend integrations to execute complex tasks as well.
For example, a chatbot on a bank’s website might provide a link to information on opening a new checking account and perhaps connect them with a live agent to set one up. A digital colleague, on the other hand, can independently help a customer open up the new account without any human intermediation.
As these four areas demonstrate, enterprises that want to build true business value through cognitive automation should look beyond low-level chatbots and move further up the AI value chain to robust digital colleagues.