Bot Really, Folks: An AI Analysis of Greater Cincinnati

I asked ChatGPT about Cincinnati and Northern Kentucky. It didn’t go well.

Illustration by Ben Kirchner

It’s too late. You’ve surely heard about ChatGPT and artificial intelligence (the abbreviation AI is everywhere now), but if you haven’t played around with a chatbot yet, well, sorry, you’ve missed the bot. The Wild West era has ended.

Having witnessed various AI platforms demonstrate that they will, without warning, start to blurt out terrifying psychopathic intentions, humanity has hastily performed several emergency digital lobotomies. We’re safe for now. The Terminator movie franchise will not come true; a Schwarzenegger-like unit will not appear. You can relax, even if your name is Sarah Connor.

Some problems remain. ChatGPT, the first consumer-grade AI, still warns on its home page that it “may occasionally produce incorrect information,” but that’s a gross understatement. It should say “almost always produces something that will make you spit out your coffee.” I’ve been giving it softball tasks, like, “List 20 famous people who were born in Greater Cincinnati.” Easy, right? Apparently not. More than half of its answers included Roseanne Barr (born in Salt Lake City), Francis Ford Coppola (Detroit), Karen Carpenter (New Haven, Connecticut), and Neil Armstrong (Wapakoneta, which at least is in Ohio).

It would seem that this impressive artificial intelligence has never heard of Pete Rose, Bootsy Collins, Doris Day, William Howard Taft, any Isley brother, Steven Spielberg, or even Charles Manson.

How could a platform linked to the word “intelligence” be so stupid? I’ll explain why, but first I want you to enjoy some other “facts” that ChatGPT explained to me about Cincinnati and Northern Kentucky.

Tell me the differences between Cincinnati’s east side and west side. “The west side has a larger number of upscale neighborhoods with larger homes, while the east side has a mix of older and more modest homes. The west side is home to more upscale shopping centers and businesses, while the east side is known for its smaller, locally-owned shops and restaurants.”

Does that look exactly backwards to you? Welcome to ChatGPT. Anyone planning to lean on it for their term paper is planning to flunk. Then again, I discovered that sometimes a chatbot’s goofiness can be sort of entertaining. Here is what happened when I asked ChatGPT to be a poet, versifying about the glory days of gambling in Newport:

Gambling halls and saloons dotted the streets so wide
A hub of entertainment, where people flocked to reside.
But as the years went by, and Newport grew
The laws of the land began to ensue.
Gambling was banned, and the halls shut down
But the memory of that time still lingers around.

Not exactly Shakespeare. It inspired me, though, to ask for some Shakespeare-isms about Cincinnati. I got a tale told by an idiot: The spicy scent of chili fills the air / With hints of pork and hints of dark and fair.

Creativity is obviously another glaring chatbot weakness. Request a joke or a song lyric, and your eyes will bleed. When you ask ChatGPT to morph into Bob Dylan and write a Dylan-esque song about Graeter’s Ice Cream, the answer is blowing:

Once upon a time, you lived a life so sweet
But now you’ve got that craving for something that can’t be beat.
Oh, Graeter’s ice cream, like a rolling stone
You can’t resist it, you’re never alone.

Um, OK. Now let’s conjure up Nicki Minaj showing Graeter’s some love:

I’m a boss, a queen, a ruler of the land
And when it comes to ice cream, I’ve got a master plan.
With every lick, I feel so fly.
The smooth and creamy texture, it makes me high.

ChatGPT adds a disclaimer when it mimics a famous writer, just in case you might think Bob Dylan actually wrote the above crap. And if you’re concerned that the kids might see lyrics along the lines of Nicki Minaj (as if they haven’t already), don’t worry: The algorithm skips over any offensive source material.

Next I tested ChatGPT’s courage to take a stand on a controversial topic. This time I went straight for the jugular: Which is better, Skyline or Gold Star? “It is subjective to determine the best Cincinnati chili restaurant, as it largely depends on personal taste and preferences. Some popular and well-regarded options include Skyline Chili, Gold Star Chili, and Camp Washington Chili. It is recommended to try them out and see which one suits your taste buds the best.” That looks like a press release from the U.S. State Department trying to avoid World War III. No matter how provokingly I worded a question about Cincinnati, Northern Kentucky, or just about anything, the answer was maddeningly diplomatic and evenhanded. ChatGPT seems to have been born castrated.

So if chatbots have wobbly accuracy, zero creativity, no cojones, and can’t do math (do not multiply past three digits), why is the entire world nervous about them taking over the entire world? Because they probably will—just not as quickly or as overwhelmingly as people are predicting. Space is too limited for me to explain why a chatbot has so many pitfalls, but here’s a crude metaphor.

Imagine you’ve been beamed down to a planet by a Star Trek transporter that’s defective. It hiccups while disassembling and reassembling your atoms, leaving behind a few million of them. Oops. You look normal, but you don’t speak or move or think quite right, because you’re literally not all there. That’s sort of what a chatbot is.

An unfathomable amount of internet data has been compressed to fit onto its servers, losing billions of digital bits in the process. It may, for example, contain deep details about Cincinnati, but only blurry artifacts for nearby Hamilton. When asked something about Hamilton, it’ll go, “Southwest Ohio, close enough…” and fake its mangled data into human-looking sentences.

Remember that a chatbot’s goal is exactly what its name implies—to simulate a chat. Accuracy is secondary at best. What’s dangerous is that each response looks as confident as your phone’s auto-complete.

Remember the laughable rollout of Apple Maps? Old enough to have used Microsoft Windows 1.0? Computer software has a long history of products sent to market before being quite ready. Getting it out there was more important than getting it right. We’re in that place now with ChatGPT and its competitors. We all need to squint at everything we read, see, and hear going forward, and struggle to decide how reliable it is. A small comfort is knowing that AI learns from its mistakes. Maybe by now it even knows Pete Rose was born in Cincinnati.

I’ve shown you what chatbots can’t do, but the tasks they’re good at are jaw-droppingly impressive and will impact daily life eventually. Should you worry? Yes, if you’re a paralegal converting 20-page legal documents into bullet points. Yes, if you’re in customer service or tech support. Yes, if you’re a scriptwriter who cranks out formula plots and characters, expecting the Hallmark Channel to keep calling.

My own career as a radio DJ is long past worry. Chatbots linked to human-sounding voice simulators means that an entire day of non-existent DJs “playing” songs and “reading” news and weather can be set up by a human in minutes. The system can then update the songs and patter every day thereafter all by itself, forever. Thanks for listening!

Artificial intelligence will shrink some job sectors, but the decline will be slow—more like the career arc of Billy Idol than Vanilla Ice. For now, chatbots are middle-schoolers who can occasionally fool you into thinking they did their reading assignment. Over time some of them will eventually master the material, ace the tests, graduate with honors, and start running the world.

The effects of AI will radiate through our lives gradually, but corporations hoping to benefit from it are galloping right now at full speed. Huge dollars are being pumped into the platform. By the time you read this column, even the lobotomized ChatGPT may have crept far into our lives. There’s a chance I might have been replaced here on this page.

Is this even me you’re reading, or a bot-generated article? Well, I wouldn’t propose such an unsettling idea unless it’s really me, would I? But wait—that’s exactly what the AI wants you to think.

Today’s chatbots can’t think for themselves and definitely don’t think about you. Not yet, anyway. So you and your job can exhale while that beta Terminator is roaming elsewhere. But he’ll be back.

Facebook Comments