AI and Customer Service: Now
Are you ready to embrace our new AI robotic overlords?
I, for one, welcome any new technology that expects to improve customer service, reduce wait times and provide more personalized experiences.
Artificial intelligence will change our world and it’s impact on customer service shouldn’t be underestimated. This guide will give you the background to discuss the use of AI in customer service and prepare your team to reap the benefits.
What is AI?
Artificial Intelligence (AI) is everywhere these days. In 2016, venture capital firms invested $5 billion, in 658 companies, in AI technology. The biggest companies in the world (Alphabet, Apple and Amazon) are all investing in and acquiring AI technology. So what actually qualifies as Artificial Intelligence?
AI is the ability of machines to display human-like intelligence, where intelligence is defined as learning from past experience, recognizing trends, or displaying reason.
Artificial intelligence is not a bot that regurgitates links to help center articles. Many chatbots simply use hard coded rules and “if/then” statements to provide answers. This isn’t true intelligence. AI should learn and improve over time as it receives more input.
Here’s some example of AI that you might have already run into:
- Google trained a computer to recognize cats in photographs
- IBM’s Watson defeated two Jeopardy champions
- Snapchat recognizes faces from other objects in order to add filters
A few other definitions that might be helpful when learning about AI
Machine Learning — teaching a computer how to do something (like recognizing a series of numbers as a phone number) by giving it data sets, telling it what to look for and then allowing the machine to make new predictions from new data. The end goal is to program software that can continue learning and improving as it receives more information.
Deep Learning — a subset of machine learning that involves less “telling” the computer what to look for, and more natural learning. It’s usually based on “artificial neural networks” — essentially a machine version of how human brains learn. Eventually scientists think we can create software that works like a human brain to discover, make connections and learn without needing to be told what to look for. It requires a ton of computing power to be able to analyze and crunch the incoming data and make connections.
Natural Language Processing (NLP) — the process of translating human speech and text into actionable software commands. Most software relies on exact inputs to understand what you need. Tyops, salng adn idoims wlil sutmp msot comupetrs. But humans can read and understand a much wider range of speech because we understand the meaning behind language. Advanced NLP is essential for making interaction with bots easier and more natural…like talking to an actual human.
Chatbots — a computer program designed to simulate conversation with humans. However, not all chatbots are AI, and it’s a big distinction. Some chatbots will simply parrot back pre-programmed replies. To be qualified as AI, chatbots need to improve as they interact with users and recognize more than a few phrases.
Using AI in customer service
There’s two main ways companies are incorporating AI into their customer service strategies. One is a front-end chatbot that customers interact directly with. The other is using AI to assist customer service reps in their conversations. Both have their benefits and drawbacks.
Front-end chatbots are able to provide instant answers to common queries customers have. For example, airlines can provide up to date flight information for passengers through chat, Facebook Messenger or Twitter. For example, China Merchant Bank uses a WeChat front-end bot to handle 1.5 to 2 million customer conversations per day. Most of these inquiries are simple requests regarding balances and payments that would take longer for humans to look up. Front-end chatbots use both NLP and machine learning to improve the service they give to customers. However, while front-end chatbots are efficient at providing instant answers, they can get stumped at more complex inquiries. Chatbots also have the potential to come across as impersonal or unfeeling.
AI assisted customer service
This is where AI assisted customer service shines. Instead of the customer interacting directly with a chatbot, they talk with a real human. But as they chat, an AI is analyzing their conversation and provides the best possible answer. The customer service agent can edit and personalize the message as needed, and the AI uses machine learning to improve their recommendations. LivePerson has seen a 35% efficiency increase using a bot/agent combination over a solo agent.
Outside of supporting better conversations with users, AI is also used to identify trends in customer behaviour, suggest proactive engagement strategies and personalize marketing campaigns. It’s a super interesting and quickly advancing technology space that we don’t have enough space to go into today.
How do I prepare?
It can be incredibly overwhelming to start thinking about incorporating AI into your customer service strategy. But that doesn’t mean you can bury your head in the sand. AI technology is advancing so quickly that companies who don’t act now could be left years behind their competitors when it comes to providing a better service experience.
Oracle found that 8 out of 10 businesses have either already implemented AI, or are planning to adopt an AI customer service solution by 2020. Are you ready to start learning?
Read, a lot. Follow people on Twitter who work on AI. Experiment with solutions (more on that below). Talk about your company’s strategic goals and how AI can support them.
AI resources to check out:
Wait but Why on AI — If you haven’t looked at this site before, don’t click if you’ve got a long to-do list. It’s some of the most amazing writing on the web, and this piece on the future of AI is no exception. We warned you!
Subscribe to the AI Trends newsletter — while some of the content is more technical in nature, it does provide a good pulse on what’s going on in AI technology so you’re up to date.
Peruse this extensive list of AI and Machine Learning resources.
Understand the limits of AI
While technology is improving in leaps and bounds, there’s still limitations to what our robot overlords can do. Complex questions, emotional pleas and edge cases; these are all conversations that humans still need to be involved in. And your AI solution needs to know when to escalate to a human.
Imagine being stuck talking to a chatbot who won’t answer your question and won’t pass you off to someone who can. So frustrating! It’s just like those telephone menus where you end up pressing 0 repeatedly until the operator comes on the line (that can’t just be me?).
Chatbots are not a silver bullet. Deciding where AI will work for your customers, and where it won’t, is a big part of rolling out AI supported customer service.
Get your data in order
AI is only as good as the data it has access to. While humans can provide a second check before sharing customer data that doesn’t make sense, AI likely won’t do the same. If your data structures are out of date and chaotic, your customers won’t get the right answers and you’ll cause panic.
The same goes for any knowledge base articles. If your AI solution uses your existing help center content to provide answers, it’s limited by what you have already documented. Review documentation to make sure there’s no gaps or outdated information.
AI powered customer service solutions
Many existing companies offering AI solutions can be setup with minimal (or no) engineering resources and integrate into your existing CRM and helpdesk. Here’s five different solutions that all offer something slightly unique depending on your needs:
TrueAI — “Our product is software that provides automatic, intelligent response suggestions for customer support operators. We can integrate our suggestions with your existing CRM system with our plugin.”
DigitalGenius — Uses AI to suggest automatic responses that agents can edit as needed. Digital Genius will also predict metadata such as tags, priorities or assignments.
Init.ai — “Automate conversations, Analyze them for actionable insights, and Assist sales and support staff in communications” by suggesting responses and next steps.
Abot labs — “Abot acts as a tier-1 service agent, replying to customers instantly, answering questions and fulfilling requests” while learning how to sound more human and using your brand voice.
IBM Watson Conversation allows you to build a bot with no coding experience whatsoever. It’s also free for the first 10,000 API calls. They also offer “virtual agent” which operates as a front-line chatbot for customer service.
AI is being used in everything from recognizing human emotions, to crunching enormous amounts of consumer data, to driving extremely personalized interactions. What will we see next? Will AI force us all out of jobs?
That’s a good question. Stay tuned, we’ll dive more into the future of AI and the problems it must overcome!