What The AI Age Really Means For Customer Service

We know that artificial intelligence (AI) is already infiltrating our lives: between Amazon Echo and Google Home, AI is becoming not just a household name, but a household item.

Yet we also know that AI is still in the very early stages of what it could potentially one day become. In order to both properly understand the range of applicable use cases today as well as quell any fears about what the future of AI may hold—it’s important to distinguish fact from fiction, and comprehensively define AI as more than a buzzword.

Think of Today’s Most Accurate AI as an Obedient Child

The most mature type of AI today is based on simple rules that are hand-coded by its human creators—it understands “yes” and “no” and has limited vocabulary. Like a well-behaved child, this type of AI listens and adheres to the rules. Consider a child learning a new language: children are able to memorize a list of vocabulary well before the point when they are actually able to understand the words’ meanings, let alone use them in a fluid conversation. This comparison is not intended to oversimplify or demean, however, as this type of rules-based machine learning has proven incredibly useful across industries in masterfully automating processes.

For example, if a chatbot is programmed to ask customers if they’d like help on a certain topic and a customer says “yes”, then the chatbot responds based on a predetermined action, with little room for error. Perhaps the chatbot directs the customer to a knowledge article about the topic, whereas if the customer had said “no,” then the chatbot might ask if the customer would like to speak with an agent (if that’s what it had been explicitly programmed to do). The chatbot isn’t “thinking” but rather acting on a very narrow set of pre-programmed responses.

AI is Just Starting to Grow Up

AI is in the very early stages of being able to understand the world around it. Think of a middle school student that is learning Spanish and knows that comprender means to understand. The student can pick out forms of comprender in a sentence, and then surmise that the sentence has something to do with understanding. The full scope of the sentence remains unclear, but picking out words and ‘guessing’ the meaning about the statement is achievable.

Predictive chatbots are able to process information in a similar way. They can pick out particular words and phrases and take a “guess” at intent of the statement. Based on this somewhat educated guess, the chatbot can suggest full text excerpts that may be relevant to the previous statement. It’s like the chatbot has access to a full answer key that contains the right information somewhere, but can do little to narrow it down.

Accuracy levels for this kind of AI vary, but the idea is that the AI feature will be able to use the information collected across conversations, compound it’s learnings, and become more accurate over time. The more information, or knowledge, that is collected, the smarter AI can become.

Not Nearly Intelligent Enough to Graduate

Gifted people who are fully proficient in multiple languages allegedly have multilingual dreams. They understand colloquialisms and hidden meanings in that second (or third) language—it’s become natural for them.

But AI today, specifically the ‘Natural Language Processing’ (NLP) field, is nowhere near that level of comprehension.

Today, AI-powered chatbots in the customer service realm can send excerpts of relevant answers with a good level of accuracy. At some point in the future, these chatbots will be able to flawlessly extract the most relevant words and phrases and formulate them in such a way that addresses any issue at hand with near 100 percent accuracy. Once this technological advancement is achieved, it would be difficult to differentiate between having a conversation with a human agent and speaking with a bot.

This is the type of AI that prompts existential, ethical questions and oftentimes associated fear mongering, but this AI is a long ways off. As Amazon and Apple customers know, frustration associated with Alexa or Siri’s limited vocabulary is still an everyday occurrence. But at the same time, these AI’s are learning new skills every day.

AI is constantly being refined and advanced, and Helpshift is at the forefront of this technological innovation. With customizable, built-in chatbots powered by the latest in machine learning, Helpshift’s intelligent customer service platform is ready to take your customer support to the next level.

How to streamline and scale customer support with AI

Similar Posts