Jawing with ChatGPT

One of my favorite movies has always been Jaws. Back when I was about 13, I remember my friend Dave and I taking the bus to a theater (that was like an hour trip – I dare you to ask your kids to do that today) and went in the theater.  Yes, my daughter now knows why I like to watch Shark Movies with her.

So, the 50th anniversary of Jaws happens this summer. I couldn’t pass it up; we made a special trip with our friends Mike and Monique to watch it in IMAX.  Our seats were up front – we could very well see right up Bruce’s snout. 

Of course, no movie release these days doesn’t come with Merch.  And to that extent there is – get your mouth around this – Jaws Wine.  It’s released by Fior Di Sole winery, who has a bunch of different labels. 

My nieces pointed this one out on the family chat.  You just KNOW that I had to put out stupid comments: “It is a biting wine”, etc. 

And then I got the very stupid idea (like THAT’s a first) to see if ChatGPT could figure out the joke. It took more than one attempt, but here is our “conversation”.

So, as it is Halloween Time, I can tell you that I had these reactions: Shock, Anger, and Fear. Truthfully, far worse than anything I ever experienced with the movie. 

I shall explain – my apologies as always.

Shock:  Let’s face it – I have this entire blog to give me the opportunity to be snarky (or is that sharky?) When my family first surfaced this wine, I hit the ground punning – the wine has a real bite to it, it’s something you can sink your teeth into, that sort of thing. My wife actually referred to the Jaws Red Wine as chewy (and how exactly did I miss that one?)

I did not expect that ChatGPT would be able to find the kinds of bits and bytes to understand that any review was intended to be punny.   In that regard, it delivered. There were even a few things that I had not thought of (fins slicing water, that sort of thing).

Apparently, I can be replaced. But you already knew that.

Anger: I read the darned thing again.

I expected that ChatGPT would do a recitation of publicly available information.  For example, “You’re gonna need a bigger glass” is right on the label. So, I am thinking that ChatGPT read the label, identified the style of humor, and recited it word for word, and then rearranged it to make it seem original.

BUT…  Note a few things that I cannot find:

  • A Blend of Chardonnay and Sauvignon Blanc.  According to Shopwinedirect.com, it is 45% Sauvignon Blanc, 25% Muscat, 15% Viognier, and 15% Chenin Blanc.
  • A 93 rating. Apparently, ChatGPT has no taste after all.

And that got me mad.  We opened the bottle, and for $13 it was ok at best.  This was never a Chard/Sauv Blanc wine; it is far too sweet for that, and having seen the grapes involved only confirmed that.  And when I searched for anything that backed up ChatGPT’s description I got nothing.

First, let’s be honest – all of the various websites had to put reviews on. But the ones that I found were parroting what is on the label and provided by the winemaker.  This description is simply made up.

BUT… A 93 Rating?  Let’s be clear – no one in existence could possibly give this wine a 93 rating. That rating is usually for a wine considered as “outstanding”, meaning one that I never drink because (now follow me closely here) I’m too cheap to spend the money

And when I searched Google, Google’s AI said that “No Jaws white wine has a 93 rating from a major wine critic; the rating appears to be based on a misunderstanding of community review scored.”  I have no idea where Google got this, because Google’s AI does not provide its sources. Rather, it blames “online communities like Vivino or WineBerserkers”.

Really, I did not make this up. 

Fear: After all of this soaked in, I found myself actually becoming a bit afraid of what is actually happening here.  I started thinking about it and have these questions:

  1. What passes as Information?
  2. What is the Truth?
  3. Most of all, where are the barriers to protect people from Siri?

Information: Trying to track down where ChatGPT got its information was damned near impossible, as it does not list any sources. 

Also, by not citing the sources, any professional sources that are being used to create these citations are not being paid for their work.  Imagine if Jaws wine replaced the need to have Wine Spectator, where a real person decides on a 93 rating because – now follow me here – they actually tasted the wine.

So, in a sense, all the money is flowing to the tech gods, and none to the people doing the work. With my daughter being a journalist, I can tell you that the journalism profession is being starved by these kinds of things, and worst of all, the management of most major media companies are embracing the tech at their own employee’s expense.

There’s a reason that Bezos just cut 14,000 jobs to his staff, to replace them with AI bots. And it’s not a good one.

Truth:   Face it – we all now rely on the tech gods for information.  And we treat it as reliable. 

Consider this – Google AI stated that the 93 rating was bogus because it came from unreliable sources (and then tried to guess who they were. Classy, guys.) As much as I would like to have Google and ChatGPT in a cage fight where they can megabyte each other like Mike Tyson, the fact that these can’t agree with one another tells you something.

But consider more meaty material. If they are relying on published material, from spurious sources, and then recycling it into other posted material, at some point all the material is, well, exactly what?

I remember my undergrad days. In one of my math/computer classes, I forgot which (maybe I should ChatGPT myself here), they talked about how errors introduced into a system can compound and blow out of proportion exponentially.  You know, like Jaws 3 (where the shark apparently can board a commercial airline to Florida for the purpose of eating Michael Caine.  Really, this is one of the best movies EVER MADE.)

And as I just typed this, I wonder if ChatGPT will read this one and pull from it to even more spurious conclusions.  Hmmm…..

Protecting People: Has the Apocalypse Come?

One of the maxims I live by is “There is value in inefficiency”.  For example, if there was no AI, 14,000 people would still have their job at Amazon. 

In our rush to embrace AI, we have put no barriers on anything that I can see.

  • No restrictions on building such as massive energy use that we end up paying for by building new power plants.
  • No restrictions on content. There was an article saying that because 80% of kids have AI do their homework for them, so rather than fixing that problem they suggest no homework.  As a volunteer for Homework Club, I say that this makes kids dumber – they need to learn and have, follow me here, Actual Intelligence.
  • No restrictions on privacy. The new Sora video app (where you can post pictures of your friends on fake videos in seconds) went out with watermarks; a few days later came an app to remove these same protective marks.)

All of that is from reading the morning paper over the last few weeks.

Enter the Terminator: We always pose the end of the world as some massive cataclysmic event headed by Arnold Schwarzenegger hunting people down one by one. It makes for great spectacle. But the truth is that things go extinct not by a massive event, but by a slow decline.  We don’t go out with a bang, we go out with a whimper.

The one truth is that we all (including myself) want to have convenience (why go to the store if Amazon delivers?) and cost (can I get it $2 cheaper, even if 14,000 other people lose their jobs?).  At some point, this all seems like we’re in a vortex circling the drain, spending our way to poverty while the tech gods laugh all the way to the bank.

Personal Truth to Hypocrisy here – I am guilty of all of this as well. I know this, and I try to counter it, but it is useless. Stores no longer stock goods on shelves – they tell you to go to the website; like you, I do. And I used Google searches a lot here to figure out what is going on.  Knowing that I am complicit is not a comfort or an excuse for being complicit here. Just something to think about. 

Most of all, I know that I provide no new insight into what has been said and that we have all heard. Like ChatGPT, I’m recirculating existing thoughts, but with my own processes (and we all know I have no AI).

So, in reflection, I pose this maxim, courtesy of Jeff Goldblum in Jurassic Park: “Just because you can do something does not mean that you should do something.”

I said this to our city council back in 2023in one of their great and silly follies.  I say to you as I said to them: “The law of unintended consequences bears this simple question: Where does this stop?”. We have created the monster; can we control it?

For now, I’ll wait for Jurassic Park Wine.  It might take a bite out of the competition.

Post-Script:

After writing this, I thought to myself – would ChatGPT even recognize my blog and possibly sample this article?  So, I put in “Review LarryLand Blog”.

It gave me a B+.  I think they’re on to me….

Leave a comment