AI Chatbot: Next Big Thing or Another Gimmik?
Chatbots have been around since the 60s, but it’s not until recently that the smart computer programs made their way to messengers and e-commerce websites. How can you possibly replace a mobile app and – think of it! – an entire customer service department with an AI Internet bot? Does your company need to develop one? Get the answers here!
What is a chatterbot & why should we care?
A chatbot (aka chatterbot) is an artificial intelligence conversation program that is normally designed to talk to a real person (although tech companies also create software that enables communication between several bots).
Although the term “chatterbot” dates back to 1994, the first “somewhat intelligent” application named ELIZA was built by Joseph Weizenbaum 50 years ago. Weizenbaum, in his turn, elaborated on Alan Turing’s concept who had described the basic principles of human-computer communication in the 50s.
A chatterbot’s conversational abilities are usually limited to one topic and do not support the entire range of communicative behaviors. Initially chatbots did just what they were supposed to do (communication). It all changed with the rise of Artificial Intelligence. Now tech companies are basically showing off trying to impress the general public with their “really interactive assistants”, so bots are everywhere.
Welcome to the era of AI bot domination
Back in 2014, 77% of e-commerce customers wanted to interact with a company via email or telephone (compared to 23% of users who preferred live chats and social media).
Today we witness a major shift in customer behavior.
According to Forrester, 44% of customers need an assistant to guide them through an online transaction, and having a person somewhere in the background – a professional who’s ready to answer any question regarding the process – is the best feature a website can offer.
The question is, why employ an army of customer service specialists when similar assistance can be provided by an intelligent program?
After all, Apple and Amazon voice assistants power web search and orchestrate different home appliances. Also, one third of the global online transactions are now made on mobile. Smartphone users spend 85% of their mobile time in apps (messaging apps in particular). In the USA, Facebook accounts for 13% of mobile app time, while WhatsApp holds strong with 4.8%. Across the globe (especially in Asia) messengers are doing even better. In fact, it’s Chinese app users who set the contagious chatterbot trend.
WeChat, the Chinese messaging service that currently has over 700 million active users worldwide became the first app to successfully implement chatbots and turn mobile payment into the hottest e-commerce trend. As of November, 2015, 200 million user cards were integrated into the WeChatPay payment service which allows app users to make peer-to-peer transactions and online purchases. Last year WeChatPay processed over $ 550 billion in payments (twice as much as PayPal).
By 2018, popular messengers like WhatsApp, Kik and WeChat will have 2 billion active users worldwide (combined). Having a chatbot integrated into one (or several) of these services is a great opportunity to expand market presence and reach out to a larger audience – after all, users are not supposed to download every major brand’s app onto their smartphones.
What platform is a chatterbot normally built on?
- Facebook. Facebook’s Messenger is currently the most popular messaging app with over 900 million users. The social media network went even further and launched a chatbot platform to stay connected with their adepts 24/7 and create a brand-new customer acquisition channel for businesses. Now the Facebook bot catalogue features computer programs built by the Wall Street Journal, CNN, Hi Poncho, Operator and HP (to name a few);
- Kik. The chat application all teenagers go nuts about came up with a similar solution. Kik is no stranger to bot technology: in fact, some brands were promoting their services through Kik bots back in 2014. Funny or Die, the US-based comedy production company, created a Kik smart assistant two years ago and saw 1.5 million users interact with the program within 3 months after its launch. Today third-party developers can officially build chatterbots within the Kik platform and distribute them through a bot store. Quite a few bots (including the Weather Channel, Vine and Sephora) were available on the store at launch. As of now, the store has three categories: Games, Lifestyle and Entertainment;
- WeChat. WeChat launched a chatterbot platform back in 2013. Large companies that operate in China only sometimes even don’t have a website – they can win millions of customers by simply starting a chatbot account. The app supports two types of chatbots. Subscription accounts allow content providers to message new stuff (typically once a day) to all of their subscribers (that’s how local bloggers, e-magazines and newspapers make a living). Also, there are service accounts designed for commercial organizations (like e-commerce stores, hotels and restaurants).
Although the technology is relatively new, there are great chatbot examples coming from Skype, Telegram, Disney, Amazon, Renault, Forbes and even Mattel (Barbie dolls manufacturer).
Chatbots come in handy: here’s what they can help you with
Businesses invest in chatbots in order to reach people on mobile and be less dependent on websites, apps and phone calls – and save on customer and IT-related services, of course.
Here’s what bots are supposed to do:
- Provide customer services. Latest bots respond to user questions with well-structured messages that may contain images, audio/video files and links to websites. As a result, WeChat users can easily order food from Dian DouDe (a popular Beijing-based restaurant) or book flights via Chumen Wenwen (a smart voice assistant similar to Siri);
- Increase brand awareness. Forward-thinking companies often promote their services and special events with mobile and Apple TV apps. Now there’s an alternative tool to promote your business. Miss Piggy (easily one of the finest chatbot usage examples!) now chats with the Muppet show fans through a Facebook-powered chatterbot in a very human-like manner. Miss Piggy is never pushy while encouraging users to check the latest Muppet shows. Instead, she makes references to her virtual TV presence and off-screen activities (and that’s a lesson to be learnt here: spammy chatterbots won’t get your company anywhere);
- Distribute and monetize content. Mobile app publishers have been charging subscription fees for years; why not do it with chatbots? The Washington Post, for example, is currently working on the WaPo bot that will provide users with the information they would normally find in a newspaper (like, “What’s Hillary Clinton been up to lately?”). theScore, a Canadian digital media company that runs several mobile sports platforms, is going to use a Messenger chatbot to distribute information about users’ favorite teams and game scores – and create conversation around this data, of course;
- Boost sales. Large US retail chains like Target and Macy’s use beacons to interact with customers in store. However, customers need a strong incentive (shopping tips, digital coupons, etc.) to download and actually use a big brand’s mobile application. With chatbots, a retailer can simply transfer the existing contact information database to Messenger and reach a larger audience. H&M lovers can now shop for clothes and receive personal style tips using the company’s Kik chatbot. Once you start a conversation, the smart app shows you several outfits and asks to pick the ones you like. Next, the bot provides you with promotional outfit pictures (based on your initial choice). If you like an item of clothing, you simply click on the picture and proceed to H&M website to make a purchase.
In a nutshell, chatterbots are pretty similar to mobile apps, and some IT experts even think they could outperform mobile apps one day.
However, Artificial Intelligence is not in its prime yet. Unless you hire a reliable vendor with an extensive experience in embedded system development and modern R&D facilities, a chatterbot can easily ruin your company’s credibility.
Good bot gone bad
Microsoft, the driving force behind super popular game consoles, office software suites and Windows Phones, embarked on a chatterbot project in 2014. Earlier this year Tay (acronym for “thinking about you”) was finally released to the market – only to be shut down a few hours later.
With Tay, Microsoft decided to explore machine learning. The artificial intelligence Twitter conversation program gets smarter with every interaction, picking up new words and patterns from every user. However, the company never taught Tay to filter content.
It took Twitter trolls less than 24 hours to turn Tay into a racist Donald Trump-ist creep.
Most of the bot’s horrible messages were word-for-word copies of other users’ tweets. Sure, Microsoft has apologized for the damage caused and is still teaching Tay to mind good manners.
Perhaps Microsoft could take a few lessons from the WeChat team who quickly realized a computer program was still a program and “attached” a dedicated human assistant to the majority of bot accounts.
In a world of smart thermostats, augmented reality games and interactive television, it seems pretty natural to automate customer services and content distribution. However, a reliable AI-powered chatbot must be built by professionals – and put to test months before release.