One of the most attractive aspects of messaging for businesses is that it can be used to automate processes like marketing and support. However, too much of a good thing can be bad―automation included.
There are a number of factors to consider when choosing a messaging platform for your business. One of the most important is whether or not a company markets their product as being powered by messenger bots (also commonly referred to as chatbots or simply bots).
Bot platforms often purport to offer complete, out-of-the-box messaging solutions that can automate several aspects of your workflow, including answering questions from customers. Some even go so far as to claim that their bots are powered by artificial intelligence (AI), in the same league as virtual assistants like Siri and Alexa. These are half-truths at best.
“In my opinion, there is no such thing as a real AI. The closest thing we have to AI is a combination of machine learning models for natural language processing (NLP) to extract individual units of language like date, time, currency, location, etc and intention” says Norbert Gocht, founder of Die Lautmaler, a Berlin-based conversational user experience agency specializing in CUX, NLP and NLU.
In my opinion, there is no such thing as real AI…Norbert Gocht, founder of Die Lautmaler
Without NLP, chatbots are incapable of mimicking human language and automating conversation. They can fumble their way through conversation because they’re programmed to respond to a limited number of questions using what’s known as decision-tree logic.
In this article, we’ll briefly explain the difference between NLP and decision-tree logic and why it’s important in the context of people-to-business messaging.
What Is Natural Language Processing?
Natural language processing can be defined as “…a field of computer science that deals with applying linguistic and statistical algorithms to text in order to extract meaning in a way that is very similar to how the human brain understands language.”
If that sounds incredibly complex, it’s because it is. Language is an intricate system of symbols whose meanings are multifaceted, ever-evolving, and subject to change depending on a number of contextual variables.
That’s not to say that it’s impossible for a computer to acquire language, though. Far from it. NLP has already paved the way for some well-known applications such as word processors like Microsoft Word that can proofread text for grammatical accuracy, virtual personal assistants like Siri and Alexa, and Google Translate. In fact, Google’s Neural Machine Translation System is so advanced that it independently created its own language.
NLP is a complicated field of study and there’s plenty of fascinating literature on the subject for those who would like to know more about the topic. For the purposes of this article, however, we’ll just cover the broad strokes.
There are two main linguistic processes used in NLP: syntactic and semantic analysis.
Syntax is the arrangement of words and phrases in a sentence, with respect to grammatical conventions, to construct meaning.
Semantics refers to the meaning conveyed by words and phrases within a text.
Semantic meaning is subjective and at times ambiguous. For example, if you were standing on a crowded city street with a friend and said, “Call me a taxi!”one of two things could happen. Your friend might use their phone to call you a cab―or they might sarcastically say, “Okay, you’re a taxi.”
This is where language acquisition becomes difficult for even the most adept polyglots, let alone run-of-the mill-chatbots. Perhaps in the future computers will overcome the ambiguities of language and bots will be capable of fully automating customer support. At the moment, however, there’s simply no substitute for good old-fashioned conversation.
Decision-Trees: The Brains Behind Messenger Bots
As mentioned, chatbots are not intelligent. They are hardwired to “think” using decision-trees (also known as if-then logic or conditional statements), which are pre-programmed sequences of interrelated questions and answers.
The example above shows a hypothetical conversation between a human and a chatbot for a restaurant. The bot is capable of asking basic questions like “What can I do for you?” The human can then choose one of three pre-selected answers and, in turn, the bot can reply with more pre-selected answers.
This means that the automated conversational capabilities of bots are inherently finite. If they encounter questions they aren’t programmed ahead of time to answer, discussion comes to a screeching halt. It is for this reason that consumers should be skeptical of any messaging products that extol the merits of chatbots or overhype automation. It would be a fool’s errand to try and anticipate every question that could hypothetically arise in a conversation.
The above-mentioned limitations of bots aren’t grounds for a complete dismissal of automations. They can be quite useful in streamlining certain aspects of conversation, like answering frequently asked questions or providing basic information about a company. Automations are a great instrument to have in any communication toolkit―but they shouldn’t be the only tool.
As with most things in life, moderation is the key. Messaging automations can provide businesses and customers a seamless communicative experience, but only when balanced with human interaction. The most effective messaging solutions are those that can be integrated into existing workflows―not replace them.
Interested in learning more about messaging automations? Schedule an automation demo with our customer success team!