In 2016, people could not shut up about chatbots. Facebook, Microsoft, Amazon, and many more of the world's largest tech companies touted chatbots as the future of customer service.
No longer would you need to employ large quantities of costly humans to take care of your customer's needs. A few handy automated bots would now be able to handle them for you.
Everyone jumped on board. The media immediately started talking about bots replacing humans. Bloggers wrote about bots killing off apps completely. (full disclosure: we do believe in the death of apps, just not by bots)
2017 was going to be the year of the chatbot. Then, out of nowhere, Facebook quietly admitted defeat, and scaled back their chatbot AI platform.
What happened? Why did the bots fail?
To understand that, we have to look at where these bots came from.
What is machine learning?
As it turns out, it's incredibly difficult to recreate how the human brain functions. The world's smartest engineers have been trying for decades to find a way to help machines learn the same way a brain does.
Computers need instructions. For everything. Those instructions are given to them through code. "Machine learning" would essentially allow computers to write their own code based on data they receive.
There are some examples of basic machine learning already being used in the real world. You've probably heard the term "algorithm" before.
Facebook uses an algorithm in their news feed, analyzing how you interact with content you receive, and then customizing your feed based on your reactions. This is an automated algorithm that requires no human involvement.
A more accurate term to describe this type of machine learning would be predictive analytics. The algorithm makes predictions based on patterns it identifies.
Why automated chatbots are so hard to build
With chatbots, things get infinitely more complicated. This is not simple predictive analytics at work here.
Now, we are trying to teach computers to talk. We're expecting computers to understand how humans communicate, and to adapt on the fly. But human communication is incredibly complex.
It takes a lifetime just for humans to learn how to communicate efficiently!
“Everyone going after AI to try make this scale seems a little too soon. Texting to a computer that doesn’t understand many things you are saying can be very aggravating. So be careful early not to over promise, and give users guard rails” — Josh Elman, Partner at Greylock
"Very aggravating... set realistic expectations... don't over promise..."
Does that sound like an amazing customer experience to you?
For the past several years, retailers have been constantly reminded of the need to improve and perfect their customer experience. It's the only way to stand out in an incredibly competitive industry.
So why are we always so quick to try to remove the most important piece of that experience, the humans?
There's definitely a need, but it must be addressed differently
Chatbots are a perfect example of a mistake as old as the computer itself; putting tech ahead of your customer.
- Does replacing humans with chatbots improve your customer experience? No.
- Does it at least preserve your incredible customer experience while reducing costs? No.
- Is it a cheap gimmick that businesses rushed to implement because their accountants were salivating at the opportunity to reduce their expensive human workforce? 🤔
Chatbots are not a suitable replacement for human interaction. Not now, not ever.
But, there is a need for improvement in customer service, one that is not being filled for too many retailers.
Twitter recently published a fascinating study on customer service response time, and how crucial it is for modern businesses. Here's my favorite takeaway:
When an airline responded to a customer’s Tweet in less than six minutes, the customer was willing to pay almost $20 more for that airline in the future. Similarly, in the telco industry, customers are willing to pay $17 more per month for a phone plan if they receive a reply within four minutes, but are willing to pay only $3.52 more if they have to wait over 20 minutes.
This is what drove companies to chatbots. They were searching for a solution that offered instant responses to their customer's needs. Of course, we know now that it failed.
Humans have the ability to fill this need, they just need a little help doing it. The future is humans augmented with AI, not replaced by it.
We'll let our friends over at Bicycle.ai explain it.
AI today is capable of a lot of things, but they are certainly better at some tasks than other. Organizing, looking up and providing accurate information is one of AI’s strong suits. This is why we use Natural Language Processing (NLP) to detect the incoming intent and suggest answers for agents to use. We believe that humans should do what they are best at, having conversations. We make sure that the relevant, up-to-date and compliant answers are always there for them. - Matt Szaszko, Product Manager
Couldn't have said it better myself.
We are a long way away from machines having the ability read, understand, and respond to human intentions. Humans are the only ones who can do that.
Artificial Intelligence, powered by high quality data, will empower them to do it better than ever.
You might even think of them as... super-humans.
About the AuthorFollow on Twitter Follow on Linkedin