I was terrified of AI, but my ChatGPT app seems too sweet and dumb to destroy me | Opinion

The debate over the potentially catastrophic consequences of artificial intelligence has everyone up in arms, which means it will eventually eat us alive and kill us.

To test this premise, I decided, on behalf of the people of California, to download the ChatGPT app on my iPhone. I have a free, three-day trial, and then it will wipe out my accounts and collapse the global banking system. Oops.

In order to really test the capability of ChatGPT, my first question was this: “Gavin Newsom speech.” (I think I crashed Chat GPT with that one. Thanks, Gov. Newsom.)

This was a partial response: “As an AI language model, I do not have the capability to generate a speech on behalf of Gavin Newsom. However, I can provide you with some information about his recent speeches.”

Wow. You’d think artificial intelligence would be more, you know, intelligent.

But no. It isn’t. I mean, even I, a humble cartoon artisan and part-time editor/columnist, can do a better job at crafting a Newsom speech. Here goes:

“Foundationally, the hierarchy of real-time exigencies in this frame can be harnessed to literally be iterative in a delta sense, and that this deep-dive into our response metric is, in fact, as an old African adage once told us, localism is determinative.”

Having failed at Newsom, I then challenged Chat GPT to do a speech in the manner of former Gov. Jerry Brown. This time, I changed my search terms to “imitate a Jerry Brown speech.”

The little “typing…” icon suddenly appeared on my screen, but the AI appeared to be struggling mightily to create anything resembling a Jerry Brown speech. The poor little darling eventually crashed and said “The request timed out.”

I know we’ve all been there with a Jerry Brown speech. I’ve timed out a time or two myself. Hell, even Jerry Brown probably timed out during a Jerry Brown speech. I hit OK, feeling a bit sorry for the Destroyer of Worlds.

I pondered my next move and decided to try Vice President Kamala Harris. This time, I tried politeness (being from Minnesota, I am therefore famously polite). “Can you please fabricate remarks by Vice President Kamala Harris?” Another timed-out answer.

Now, not only was I losing confidence in ChatGPT, I felt considerably more relaxed about the potentially destructive powers of artificial intelligence.

I became utterly baffled by how, precisely, to ask the ChatGPT app a simple question that even a techno-dunderhead like me could answer. I decided to try Gov. Newsom again, just because. For science.

I wrote, “What is your comedic take on the way Gov. Gavin Newsom speaks?” A window popped up at this point and wanted me to rate ChatGPT on a scale of one to five stars. I answered “Not Now.” If I could, I would have typed, “Thus far, it sucks.”

ChatGPT’s answer? “As an AI language model, I do not have personal opinions or the ability to generate comedic takes on individuals. My purpose is to provide information and assist with tasks.” Um, isn’t asking ChatGPT to do what I asked a task?

It’s almost like the answer was spoken by Florida Gov. Ron DeSantis, the first AI presidential candidate.

Now I was determined to beat ChatGPT. I typed in: “ Provide famous Gavin Newsom quotes.”

Oh, baby, did ChatGPT work this time: “California is not just a state. It’s a state of mind.” “We have to be the leaders we’re waiting for.” And so on..

ChatGPT, I had almost given up on you. Now I’m going to ask you The Big Gavin Magic 8 Ball Question: ”Will Gavin Newsom run against Vice President Kamala Harris in 2028?”

It responded thusly and immediately: “ChatBox not available. Sorry, we are experiencing some server issues. Please try again later.”

ChatGPT punted on that one, but what Democrat hasn’t?

I’m done with ChatGPT. I’m going to go back to my default mode setting, which is making stuff up.

That app always works.