Over the weekend I tried to catch a Lyft. He wasn’t where the app said he was, and when I called him, 3 times, he hung up on me. I ended up canceling that Lyft and ordering another one, but I was upset to find out that I was being charged $6 for the canceled Lyft. While in the Lyft that picked me up, I took to the app to complain. I was sent to Lyft’s help bot, and after clicking a couple of buttons and only sending one message summing up my experience with the previous Lyft I was issued a refund immediately. Without ever speaking to a customer service representative and taking much less time than trying to find the Lyft that never was.
The Lyft’s chatbot was remarkably well done and was able to turn my unhappy experience around quite quickly. In way less then the time that I’m ever on hold to get a real human being on a phone, my issue was completely resolved. It used some pre-selected responses to help speed up the process and increase accuracy.
Chatbots are great for this type of customer service. Especially with a service that is already built around technology and your phone. Being able to figure out what a customer wants/needs and deliver that service right away can speed up the process and keep the customer happy when done right. (When done wrong, it wastes the customer’s time and just makes them even more upset.)
Because of advances in chatbot AI, they are becoming more and more common. Many websites now have one to try to help you navigate issues, or assist you in picking out something to purchase. As of 2017 4% of companies were using chatbots, and 80% were planning on using them by 2020.
Chatbots use Natural Language Processing (NLP) to figure out what was said and to come up with an appropriate response. Some use Natural Language Understanding (NLU) which is a subset of NLP that deals with machine reading comprehension. It takes unstructured text and tries to classify it into structured entities for better comprehension. They can also use Natural Language Generation (NLG) to turn structured data into text. NLU looks at the input where NLG generates the output.
Today it’s very easy for someone with little technical experience to build their bot. Services like Amazon’s Lex use a pre-existing chatbot (this one uses the one that Amazon built for Alexa) and let the customer add their unique parameters.
Chat services like Slack and Discord even have their built-in bots and bot customizations that people can set up to perform different functions, such as sending out automatic reminders, responding to specific words, or even tell jokes.
Chatbots are here and they are fast becoming ubiquitous for more and more tasks. They have the potential to revolutionize customer service and how we interact with our computers. They also have the potential to be a huge cost saving for many companies. It’s not news to the tech world that we are going to see a lot more of them in the future, with the ease of implementation I hope that I no longer have to waste time on hold to deal with a simple customer service issues.