Why Aren’t Chatbot Merchants Using Chatbots Themselves?
Chatbots, says “online-marketing professional” Murray Newlands, are “extremely sophisticated and versatile tools that can help you automate a number of your business processes.” Newlands goes on to explain that “AI-powered chatbots can be part of your marketing initiatives and help push your customers seamlessly through your sales funnel.”
Golly, if chatbots can do all that, who wouldn’t want to start using a chatbot? For that matter, who hasn’t dreamed of being pushed by a robot through a sales funnel?
Murray Newlands isn’t just a disinterested “online-marketing professional”, of course. He has a chatbot business of his own, convincing people to pay him to help them set up chatbots for their businesses, “to push your offer or deal to users on demand”.
He calls it ChattyPeople, although with chatbots, it’s not people you “chat” with – just software routines… Or is it?
I went over to ChattyPeople to see just what the chatbot expertise of Murray Newlands could do for me. The first thing that I discovered is that there are actually no chatbots at ChattyPeople to help you find out about how to get started with the site’s chatbot services.
To start the process of learning how to buy chatbot services ChattyPeople, you can only read non-interactive blocks of text on the screen, or fill out an old-fashioned web form to “Request a call”. Yes, the power of chatbots to push people through sales funnels is so astonishingly strong that the Murray Newlands and his minions at ChattyPeople would rather that you just let them place a plain old telephone call to you instead.
Murray Newlands has come up with plenty of dubious ideas before, such as when he asked his followers on Twitter to “Be a voice, not an echo,” and many of his followers agreed – by retweeting the post. Promoting chatbots as a get-rich-quick marketing scheme, but then neglecting to use chatbots yourself, though, is a new level of chutzpah.
To be fair, Newlands isn’t alone in this scheme.
Tywla Helps, another chatbot company, doesn’t have any chatbots on its web site either. Like ChattyPeople, Twyla Helps asks people to fill out a web form, or send an email instead.
Pandorabots asked that I call a phone number, but only after 10:00 AM Pacific Time, which is 1:00 PM Eastern Time, and the middle of the night in the rest of the world.
I looked and looked and looked, but I couldn’t find a single business that sells chatbots that actually uses chatbots to communicate with its customers.
It turns out there’s a reason chatbot companies don’t use chatbots to chat with their customers: Chatbots don’t really work very well.
I went to the self-proclaimed top experts in artificial intelligence at IBM Watson, to see what that company’s showpiece chatbot could accomplish. The results weren’t very promising.
I asked the chatbot, which took the form of a driving advisor, a simple, appropriate question that any human marketer would be prepared to respond to. “Tell me why I should use a chatbot,” I wrote.
IBM Watson’s supercomputer didn’t know what to do, so it talked about itself instead. “I’m trained to turn things on or off in the car dashboard, play music and find nearby amenities like restaurants, gas stations and restrooms,” it responded.
I persisted. “But why should I use a chatbot instead of doing it myself?” I asked.
IBM Watson seemed determined to let me know that the conversation was not about my questions. “I’m trained to turn things on or off in the car dashboard, play music and find nearby amenities like restaurants, gas stations and restrooms,” the chatbot said.
Would IBM Watson give this response to anything I said? I asked, “Why should I ride a skateboard?”
Once again, IBM Watson spat its default message at me. “I’m trained to turn things on or off in the car dashboard, play music and find nearby amenities like restaurants, gas stations and restrooms,”
This isn’t chatting, I thought. I decided to get more philosophical with a pop culture reference. “Why do birds suddenly appear every time you are near?” I asked.
IBM couldn’t tell I was coming on to it, apparently. It merely noticed that I had used the word “time”, and so responded, “My system tells me that the date is Tue, 5 Sep 2017.” That wasn’t even the time of day.
“Did you really think I was asking about the time?” I wrote. “Doesn’t IBM Watson have popular song lyrics in its database?”
Once again, IBM Watson recognized one word that it knew. It decided that I had given it an order. “Great choice! Playing some pop for you.”
I decided to provide some simple correction to the chatbot. “I didn’t ask you to play music for me,” I wrote.
IBM Watson was unperturbed, and started making judgments about my likely music preferences. It responded, “Unfortunately I don’t have any Soundtrack music in my collection. Try genres like jazz or rock.”
I couldn’t take it any more. If IBM wanted merely to respond to driving directions, I would give it one. “Drive off the road and crash,” I typed into the driving simulator.
Even this message was not clear. “I understand you want me to turn on something. You can say turn on the wipers or switch on the lights.”
Off. I said the word “off,” and Watson thought I wanted something turned on.
Somebody call Ray Kurzweil and tell him that he might want to postpone making any more predictions about The Singularity for another 25 years or so.