My recent experiences with artificial intelligence-powered chatbots, especially those used for customer support, have left me feeling irritated rather than amazed — and I don't think I'm alone.
How do I hate chatbots? Let me count the ways:
Chatbots Want to Talk Your Ear Off
If I chat with a human being, I can express complex issues and speak in long, content-rich paragraphs. The person is able to decode and parse what I say, even when there is more than one thought involved. This makes my interaction more efficient.
Chatbots, on the other hand, aren’t able to consume a firehose of human language. They keep coming back and asking question after question. It’s like a chatty neighbor who keeps talking and talking when you have work to get back to. With chatbots, there is a lot of repeating and “did you mean this?” chatter. They have to be walked through a problem while a person can hear the whole problem and break it down themselves.
I ran into this with Microsoft support. I needed to ask what seemed like a simple question about my Microsoft 365 account. It was the type of question a human could (and did) answer in less than a minute. The Microsoft chatbot, however, couldn’t understand my question and kept responding with more and more irrelevant questions of its own in a vain attempt to understand what I wanted. I finally had to ask for a human.
Related Article: A Good Chatbot Is Hard to Find
Chatbots Take More Time Than a Human
Since chatbots are so inefficient in their “speech” they take a lot longer to find a solution outside the most common ones. That would be only mildly annoying if they offered a single interaction such as “What can I help you with?” followed by a “Is this what you are looking for?” answer and then a pivot to a human agent if it didn't resolve your issue.
Unfortunately, that’s not how they work. Instead, they continue to ask questions until you give up. Sometimes, they don’t give up even when you demand a human. I recently went to the Verizon Wireless customer service chat expecting a human. It was obvious I was speaking to a chatbot — it kept making really silly suggestions — and I demanded an agent. The chatbot persisted until I asked for an agent three times and then it sent me into the chat queue. The question was more complex than the chatbot could handle but it wouldn’t give up. The chatbot wasted my time before finally getting me to an agent that could help … two minutes later.
It’s Not About Service — It’s About Costs
Chatbots leave one with the impression that chatbots are not intended to front-end a human interaction or gather information to make that interaction more meaningful. Instead, it is pretty obvious that chatbots exist to keep you from speaking to a human. They are just like interactive voice responses (IVR), which also exist to save costs by keeping you from a human, only they require more effort. It’s much easier to say a simple command or press “1” on your telephone keypad (although, in all fairness, I do find myself yelling “Agent!” at the IVR voice recognition a lot. So I guess I hate IVRs as much as chatbots.)
What makes the cost shift obvious is the implementation of most customer service chatbots. They don’t ask for some basic information (account number or product model) which could then be used by a human agent. Instead, they try to solve the problem and keep doing that even when it's out of its league. This is just an IVR jail, only with more typing.
Related Article: 8 Ways to Measure Chatbot Program Success
Chatbots Yield the Same Results Any Search Engine Would, But With More Effort
If a chatbot returned superior results — if it could actually solve my problem with less effort — then it might be worth it. But for anything more complicated than a quick database retrieval, chatbots return results that any good search engine would. All that the Verizon Wireless and Microsoft chatbots did was return links to content on their customer support site that was easier to access from the sites’ search engines.
One reason a chatbot takes more effort than either self service or a human agent is because, compared to us humans, it’s stupid. A person could make the decisions necessary to obtain targeted information from website content much faster than the chatbot can.
In the end, chatbots are like swimming with clothes on. You get to where you are going, but it takes a lot more time and effort.
It’s Easier to Use an App
If the real purpose of a chatbot is to make customer service interactions easier when on a smartphone, then how is it easier than an app? The answer is it isn’t. Yes, sometimes you may not want to download an app for the occasional company interaction. It is also true that the mobile browser experience tends to be second rate. Overall, building customer service into an app makes it easier to get service by pushing buttons rather than typing away on a tiny screen. And, a lot could be accomplished by improving the mobile browser experience. Instead, we have chatbots making believe they are people and not doing a convincing job of it.
Related Article: The Humanoid Touch: How AI Is Changing the Customer Experience
No, Really, It’s About the Money
To be fair, AI-powered chatbots are new and not yet capable of truly emulating a human interaction. So, why are companies rolling out these truly inadequate chatbots? One theory says it’s a necessary step toward achieving real AI chatbots. Chatbots are powered by machine learning and need to be trained. That can only happen when you have real humans interacting with it. If this is true, then real customers are being used to teach the chatbot how to do its job. That's like telling customers they have to come in and train the customer service agent if you want them to help you. Even if it’s true, it’s bad service.
A more plausible theory is that how well chatbots perform is unimportant. Instead, chatbots are yet another attempt to shift costs from companies to customers. By forcing untrained chatbots on us, companies make us spend our time on a type of service that is both less efficient and less satisfying. Companies save money at the risk of our time and irritation.
I can see how a vendor might think this way. Customer service costs are easily quantifiable in terms of personnel, facilities and equipment. Customer time and annoyance are not as easily counted, and the effects are indirect and won’t show up on a balance sheet. Until revenue plummets that is. This was, and still is, the story of the IVR and that hasn’t gone away yet. Everyone hates IVRs, but they are ubiquitous.
Perhaps I’m just being curmudgeonly or dislike having human-like interactions with machines that lack human intelligence. Trying to fool us with machines masquerading as humans affects us at a visceral level.
Or, maybe chatbots are just not the right technology for customer service interactions. This could turn out to be just a bad experiment.