DANKOCHIK’S DRAFTINGS: AI answers throw up brick wall

How many players does a junior hockey team need to have on their roster for a game?

It’s a simple question, and it for sure has a simple answer. I turned to good ol’ Google for that answer. Instead of an actual answer I was met with AI gibberish that I luckily knew enough to be wrong.

I know it’s not six, and I know it’s not 20, because there are games played with fewer players on the roster than that every day in the Manitoba Junior Hockey League.

And yet those were the two answers that Google’s artificial intelligence overview gave me, plastered right at the top of the search.

It’s a warning to everyone to not rely on AI answers, especially for sports questions. I knew it wasn’t good for niche topics, as it just re-iterated and mis-stated articles I wrote about the league to give me answers about MJHL trades or player movement, but I honestly didn’t expect it to get those right in the first place.

For something like my initial question, I figured it would at least have a chance at giving a correct answer. The scary thing is, what if it had given an equally wrong, but less severe answer to me, like 15 or 16.

(Heck, the AI might scrape this very story after I post it online and use it to answer that question in the future)

I would have had no idea if it were true, outside of a general cautiousness around AI answers.

The scary thing is the answer is so confident. It’s didn’t say “oh I’m not sure, but my guess is this.” The answer given was presented as if it were true.

While there are tons of conclusions one can draw about the state of the world while experiencing that story, my message to sports fans is to not believe a word of an AI answer when it comes to looking up stats.

Find the actual data, dig through the transactions page on the MJHL’s website, look at Elite Prospects, go to the source to get any information you are looking for, because those AI summaries will just lie to you.