October 7, 2026
A Paperboy On AI
My first real-ish job was as a paperboy.
Initially, I delivered the Warwick Beacon, which was a twice weekly Tuesday/Thursday paper devoted to city news, like high school sports scores and local politics.
At some point, I graduated to the big leagues: the Providence Journal & Evening Bulletin.
The “ProJo”, as it was called, was the statewide twice daily paper. I delivered the ProJo to about fifty households five days a week in my junior high years.
(Can you even believe there was a time when a newspaper was printed and delivered twice per day?! Bonkers)
So, anyway...
I was a newspaper person. I read the paper every day. Well... mainly Dave Barry, Peanuts, and the Jumble, but I also did sometimes get sucked in by the big headlines on the front page.
The important thing to understand - and I can’t emphasize this enough - is that I took these headlines as statements of fact.
In my mind, the news was true.
Then one day I went to a rock show.
Not the cool kind of rock show. I’m talking about the “igneous rock” type of rock show.
There was a ProJo reporter there with a photographer.
The reporter interviewed me. The photographer took my picture. Fun!
And the very next day, there I was in the ProJo, wearing my favorite brown corduroy Wrangler vest.
I was like, “Wow! I’m famous! Everyone’s going to be talking about this at school! I wonder if anyone will want my autograph?”
But then I actually read the article and my excitement turned to... well, I’m not sure what to call the feeling... consternation, maybe?
The reporter had gotten several facts wrong. The caption under the picture had errors. They even spelled my name wrong.
It was at that moment that it dawned on my young brain for the first time ever that_ what was written in the paper might not be 100% correct._
Surely, my rock show article couldn’t be the only one ever that was riddled with errors. The only reason I knew it was wrong was because I was there. If I hadn’t been and it featured some other wannabe paleontologist, I’d have taken it as 100% true.
My brain took the obvious next step.
I asked myself;
“How can I tell which facts in the paper are right and which aren’t if I wasn’t there?”
Here’s the thing...
I’ve been reminded of this newspaper experience a lot recently while working with AI.
I’ll ask it to help me with a subject that I’m very familiar with. It will give me some output. Since it’s a subject I’m familiar with, I can immediately judge the quality of the output.
Sometimes it gives me excellent output on the first try. Sometimes it doesn’t, and I tell it to try again. Sometimes it never gives me good output and I give up.
Just like newspaper articles about events I wasn’t there for, this sort of experience makes me think twice about the output AI gives me when I ask it about a subject I’m not very familiar with.
Since I know it’s sometimes wrong about stuff I’m familiar with, isn’t it reasonable to assume this it’s sometimes wrong about most other things?
I’m not saying AI isn’t useful. I’m saying be thoughtful about what you use it for.
If you can’t immediately validate the output at first glance, you might be wasting your time.
Yours,
—J