Image above generated by DALL-E from ChatGPT

By Larry Felton Johnson

This “From the Editor” column is an expanded version of the following Facebook post I did this morning:

Before I begin, I should state that I have no problem with artificial intelligence. I use it here at the Courier, mostly for mundane tasks like alphabetizing and formatting lists of counties from the National Weather Service to make our weather reports readable, generating generic stock images like the feature image on this column, and helping me correct typos.

ChatGPT saves me from having to write computer utility programs for every frequently used text formatting task. Some days, I spend most of the day reformatting badly mangled text from various sources and making it readable.

However, there is one aspect of artificial intelligence that I do not like.

The number of AI chatbot scam phone calls has increased to the point where I get more than ten per day, every day. The nature of the scam changes from month to month, but the latest two are funeral insurance and accident claims. The particular scam doesn’t matter—a scam is a scam.

The way the chatbot works is impressive to me. It’s come a long way since I sat at an old black-and-green computer terminal in the 1980s.

I come from an IT and programming background, and there have always been text-based chatbots. You would enter a sentence and the bot would check a table of possible responses and repeat one back. A well-known one in the 1980s was the psychoanalyst program. You’d type in “I’m feeling depressed,” and the program would respond with something like “Why do you think you’re feeling depressed?”

With the current state of AI, the programs have gotten a lot more sophisticated.

An audio chatbot can make a statement, evaluate your response, and respond with an answer that guides you toward the outcome the owner of the chatbot wants, invariably to separate you and your money.

After getting dozens of these calls over a several-day period, interrupting already busy work days, I decided to have fun with the bots.

My goal was to see how confused I could make the AI scam chatbot, and to determine how many confusing loops the bot would go through before breaking off the call. I began doing it yesterday. I haven’t yet begun logging the results in any systematic way, but had a lot of fun playing around with gibberish to feed the chatbot.

“There are flying fish in my hair!”

Many AI scam chatbot robocalls begin with something like, “Hi, I’m April from scam-of-the-week. How are you doing?”

So I came up with an initial response: “There are flying fish in my hair!”

When the confused chatbot tried to continue the dialog, I said, “There’s lots of fun at Finnegan’s wake!”

Other things I said over the course of three experimental calls included, “Ninety-nine is too many bottles of beer for that wall!” and “Can you please tell me how to get rid of this hideous hobgoblin?”

I really missed my chance with one yesterday who repeatedly asked my age no matter what gibberish I babbled. If I get that one again, I’m going to say, “I’m 297 years old.”

If the scambot has a response for out-of-range numbers, I’ll say, “Hey, 297 is young for a creosote bush!”

Another list of possible responses

I’m working up a whole library of gibberish responses for the entertainment of the automated chatbots.

Here are a few off the top:

• “Would you like to hear a dramatic reading of Joyce’s Ulysses? I’ve got a copy right here: (clears throat) ‘Stately, plump Buck Mulligan came from the stairhead, bearing a bowl of lather on which a mirror and a razor lay crossed …‘”
• “Why is there tickweed in my captain’s chest?”
• “Can you yodel? I’ve always thought phone yodeling could be a great art form”

The possibilities are endless.

Of course, the sensible thing would be to just hang up. But since the calls already interrupt my work flow, I might as well take a few seconds to have some fun with it.