Whether we’re ready or not, the next wave of bots are here: chatbots. Automation at its finest, these bots serve a multitude of functions ranging from personal assistant, to chatting about the weather, to providing flight delay updates.
And the best part? The interaction takes place right on your smartphone. One more reason mobile is winning for advertisers.
So, what makes chatbots so special from other robots? Chatbots unlock “conversation as a platform” explains Microsoft CEO Satya Nadella. Essentially they’re turning a human language into a user interface.
Companies like Microsoft, Facebook, and others are jumping on the chatbot bandwagon, hoping to engage consumers with a unique form of customer service (and of course cut overhead costs, too). Before you decide to join them, here are a few things you need to know.
1. How Chatbots Work
Remember Pee-Wee Herman’s robot, Conky? Each episode, Pee-Wee would go to Conky and ask him for the ‘secret word’ of the week. (Then we would all scream, when the word was said.)
Chatbots are a modern day form of Conky. You simply request information and get a quick response in text format.
They’re intelligent enough to understand the request and provide appropriate answers and/or actions. What differentiates chatbots from apps is their ability to provide info, purchases, and services within a text format through messenger. The user doesn’t have to open and run a separate application. Eventually, chatbots may end up being integrated into mobile apps, but for now, they operate separately.
2. Types of Chatbot Platforms
Depending on the business and consumer, there are a variety of chatbot platforms out there. Here are a few popular ones:
KiK. Millennials favor Kik to chat, shop, and share content with their friends. Categories include entertainment, lifestyle, and games. You’ll find brands like Vine, H&M, and Sephora, just to name a few.
Line. Chatbots are taking the world by storm, and Japan is the latest to launch their own version: Line. Still in the testing phase, the company is only offering 10,000 trial test chatbots. Eventually, they hope to use their chatbots to automate customer service, food reservations, etc.
Telegram. Consumers looking for fast messaging need look no further than Telegram. The platform can teach, integrate with other services such as creating a tg bot for casinos, automated customer services as well.
The Chat Bot Club. Currently in development, The Chat Bot Club is working on a chatbot that will be you. Using Cisco Spark and IBM Watson, it will be able to mimic your favorite phrases, emojis, etc. The goal is to integrate APIs for WeChat, WhatsApp, Kik, etc. So instead of you trying to keep up with all of your friends, the chatbot will respond as you.
Facebook Bots on Messenger. Never one to be left out, Facebook is allowing their businesses to deliver automated customer support via chatbots, too. For instance, the platform enables consumers to order flowers from 1-800-Flowers simply by sending their Messenger bot a friend’s name. You can also book a hotel or make restaurant reservations .
If your business thrives on consumer interaction and you’re looking to save manpower, then chatbots are definitely something to consider.
3. Not All Chatbots are Completely Automated
Chatbots are a work in progress, so it’s no surprise that not all chatbots work autonomously. Think Butter’s precocious bot A.W.E.S.O.M.-O who was actually young Eric Cartman in disguise.
Source: South Park Studios
X.ai. is one such company who uses Supervised Learning to train their chatbot Amy Ingram, an AI personal assistant. Humans a.k.a. AI trainers are used to verify machine-generated annotations. They spend their days looking at fragments of text the machine has extracted, then accepting or modifying to ensure Amy understands human nuances.
AI trainers are essential to clearing up ambiguity and raising the level of accuracy. To build a truly autonomous agent, you need training data, and humans to label that training data.
X.ai. breaks the process down as this: “We needed to create a dataset from scratch, and so we quickly had to build the data-collecting machinery itself. These systems are temporary but essential. Imagine you are building the New York subway system at the turn of the 20th century. You need to dig a bunch of massive tunnels. And to do this, you first need to custom build the drill, since no such system had yet been built at scale. In 1900, you can’t buy an “off-the-shelf,” subway tunnel drill.”
They don’t trust their bots to fly solo just yet, so the humans serve as training wheels to ensure customer service remains top notch while refining the model.
4. User Reactions Are Mixed
So the million dollar question: are chatbots worth it? For many consumers the answer is no. Trying to use bots for simple tasks can be frustrating and far less efficient than simply picking up the phone and talking to a customer service rep. One user had a hard time trying to get a bot to provide options for a simple black shirt (the bot pulled white shirts instead). While others have encountered shady chatbots who simply refuse to give a straight answer.
Meet Poncho, a sarcastic weather chatbot. Normally, we appreciate sarcasm, but not when we need answers, and need them fast. Here, this person had a confusing exchange with the bot. Instead of Poncho delivering the weather report, he demanded they ask their question again in “Cat.” Talk about demeaning and bizarre.
Clearly, some chatbots have kinks that need to be worked out.
5. What Does This Mean For Customer Service Jobs?
A negative side effect of automation is loss of jobs, and chatbots are no exception to this rule. Case in point, WeChat, a chat app used to sell products, originally employed people to read messages, text back, and sell the item. Now WeChat is replacing employees with chatbots.
But until we’re confident in letting chatbots operate on their own, there will always be a need for humans to monitor their AI counterparts. Finding a healthy balance between chatbots and human employees is the key to increasing efficiency without making human workers obsolete.