While writing yesterday’s post about the demise of the last JOA, I used Google to search for some of the history. Results varied wildly, especially the AI search summary that precedes the display of individual links.
Here’s a good example.
I was looking back to 1992-1993, and trying to recall when Gannett dumped the Honolulu Star-Bulletin, which it had purchased back in 1971, and instead bought the Honolulu Advertiser, which had become the larger newspaper as American reading habits changed in favor of morning papers.
So I had a simple search query that I put to Google: “When did Gannett buy the Honolulu Advertiser?”
The response came back quickly.
Gannett didn’t buy The Honolulu Advertiser; rather, they sold it in 2010 to Black Press, which then merged it with the Honolulu Star-Bulletin to create the current Honolulu Star-Advertiser, with Gannett having previously owned both papers before the sale.
Had I been a high school student using AI to write a paper, I would have probably run with this answer and ended up in serious trouble.
But, luckily, I knew enough to know that Google’s summary was simply wrong. Period.
Today I went back and made a slight change in the search. This time around I put the question a little differently.
“When did Gannett take control of the Honolulu Advertiser?”
And this time, Google’s response got it right, contradicting its earlier answer.
Gannett took control of The Honolulu Advertiser in January 1993, purchasing it from the previous owners to keep it within its corporate fold, even as it sold the rival Star-Bulletin to maintain a Joint Operating Agreement for a few more years before the papers eventually merged in 2010.
I haven’t give any deep thought to what this means, beyond “buyer beware” when it comes to using AI to get easy answers.
Discover more from i L i n d
Subscribe to get the latest posts sent to your email.

“luckily, I knew enough to know that Google’s summary was simply wrong”
most would not know enough.
period.