A fantastic Greek place in Williamsburg double-billed me last week. These things happen. Thankfully, nowadays I find out about strange activity on my account pretty quickly, and can act on it just as fast, with no need to wait on hold. The tools engineers like us build, like the live chat option on my bank’s mobile app, can be fantastically helpful. A simple process opens up the chat screen wherever I am:
“I was double-billed by Ela Tavern and want to dispute one of the charges”
“Do you mean this charge?
Ela Tavern – 10/6 – 45.60
Ela Tavern – 10/6 – 39.72”
“Yes” (Wow, that was fast.)
“Can you tell me which charge you are disputing?”
I explained the situation, where a slight language barrier during our initial order call turned into us getting a call back, where we gave our card number again and presumably were double-billed in error. I also said I’d prefer to dispute the higher charge, but would be happy to pay either (but only one).
“Can you help me understand which charge you are disputing?”
That response surprised me. How dull. “Wait a minute, is this a bot?” I thought to myself. As someone who builds bots, it was exciting to see one in action. As Oscar Wilde might have said, the suspense was terrible and I hoped it would last.
“The 45.60 charge”
“Is there anything else I can help you with?”
“Can you pay my bill this month?” I was curious to see just how capable this bot was.
“I cannot help you with that.”
“Oh well. Can you help me with something else?”
“My card is cracking and I think I need to replace it. Can you send me a replacement card—not a new number, just a replacement”
“I can help you with that. Your new cards are in the mail. They should arrive in 7-10 business days. Is there anything else I can help you with?”
“Wow, that’s awesome. No, thanks!”
At this point multiple paragraphs of rather formal text were displayed, written in the first person. “It is my honor to have served you” (honor?), “You are a valuable asset to XXX bank” (asset?), and so on.
I told my wife about the experience: how impressive the bot’s parser was, how I used pretty natural language and it quickly picked up my intentions. I know it can be difficult to consistently get that right. She, of course, was glad our new cards were on the way.
Happy as a clam, I went to get my hair cut. David, my barber, was waiting. I told him the story.
“Yeah, I mean the whole thing with the smiley face and joke was lost on it but it knew where I was going”
“Pretty great, right? The bots are coming!”
“Just like in The Terminator”
“Yeah, I guess so”
Things got more interesting when I tried to pay for my haircut: my card was declined.
“Ha, it’s probably the cracked card. Had to happen sometime!”
“Here, try this one”
“Yeah, this one will work”
Walking out with my receipt, while trying to convince myself that it really was the physical damage—and not that stupid bot—I got a message from my wife:
“FYI: credit card was deactivated by the bot. Spoke w rep and she said chat can’t replace card only send new card w new number. She expedited the cards—should have them by Tuesday. Also giving us 1000 points”
A cautionary tale. Just that morning I was praising it, and now this bot was mucking up my life. What if I didn’t have another card? In a perfect world, AI continues to make our life better. While things have certainly improved, the process has its fair share of hiccups (and deactivated credit cards). Forget dystopian science fiction about the rise of the machines; smart AIs aren’t what we need to worry about, it’s the dumb ones we need to watch out for.