Testing out “Chat” and “Gemini” yields mixed results
A COLUMN by Steve Sprague
For a while, I was very proud of the hand-held calculator I bought as an “early adopter” in 1973. If you’re tinkering with the available Artificial Intelligence (AI) systems online today, you know the feeling.
It won’t last.
Stories show that the multiple AI platforms are now displacing actual web searches for information people think they want or need. My calculator let me think the math I struggled with in school was also now a quick and simple answer. Not long after that, I was spirited off to Washington by a new congressman. That once-marvelous calculator was pitifully useless in the face of federal budgets in the trillions.
Before Congress, as a reporter for the Star-Gazette, I would have sold my soul for the immediate information now instantly available on every computer in the country. Research, how-to instruction, recipes and background on virtually anyone. Instead, back then the process was libraries, how-took books and mostly shoe leather to personal conversations. And trust.
Now that AI has entered the picture, that last element of “trust” has yet to be determined. For now, however, my strongest advice is to treat AI’s answers as “information,” not “fact.” Especially if you’re looking for meaningful facts about people or events to reinforce arguments you want to promote.
For honest and diligent professional reporters, the standards never changed. Collect information from “sources,” but it doesn’t get into a story until it’s confirmed as “fact.” One of my favorite recollections of the difference came while at the S-G watching a local newscaster. He prefaced a story by saying it was “an unconfirmed report from a usually-unreliable source.”
Oh. So, it’s gossip?
AI can tell you lots of things. It can speed through lots of “grunt” work that might otherwise cost you hours. It can confirm your chicken needs to cook to 165 degrees. In many ways it can be a massive help for small businesses that can’t afford extra office staff. But it ain’t to be trusted as a reliable source of unconfirmed “facts.”
As a test of today’s AI, a close friend and I both asked our AI buddies to tell us what they know about me. He used ChatGPT; I used Google’s “Gemini.” What came back was the blinding light of “limited sources.” Both services spit out only information it could find on the net. Date that as only places I’ve appeared online since 2021. Mind you, these were the “free” AI searches.
My friend’s ChatGPT regurgitated my abbreviated employment history from the few references made in The Sun alongside previous columns and bits from my Linked-In page. It also cautioned the searcher that it could not find any documents confirming my claims to have worked for Congress or even my status as a veteran. “I found no mainstream news coverage of his veteran or Congressional-staff roles beyond the local opinion-piece references.”
My Gemini report had even less information and didn’t bother to remind the searcher that all this was unconfirmed. At this stage, it would seem ChatGPT was marginally more useful.
In other words, I could be the local version of ex-Congressman George Santos. In truth, there are countless public records of my income and titles while working for Uncle Sam, literally hundreds of by-lined stories written by me as a reporter and then about me in those four careers.
If you’ve followed AI to this point, you know it accumulates and recalls whatever databases it has been exposed to. Maybe, if you’re ready to pay an extra fee, at least ChatGPT said it could “attempt a deeper public-records check.” Apparently, the Star-Gazette and the now-gone Steuben Courier-Advocate don’t qualify as “mainstream” news sources. Maybe even the Wall Street Journal where I was a story source. Hard to tell.
Both my friend and I are curious but cheap. We didn’t pay to see more. But you get the point. AI today is an infant growing exponentially and, like any infant, needs careful oversight and reasoned control to grow into a responsible adult. Treat it as such.
There’s no question AI will have a more-than-significant impact on every aspect of our future, from energy demands to jobs. But there’s still a lot of snake oil consumers and gullible kids who will pay the price without meaningful guidance.
Please, do your part. Monitor (or block) AI use by kids and encourage skepticism of AI’s wonders and quick responses for your quick-to-believe friends or relatives.
Steve Sprague is Bath NY based Columnist for the Sun who addresses contemporary topics and looks back on important notions from his career in Washington DC. You can reach him anytime, sgsprague@gmail.com
