Once upon a time there was a small desert village with a single well outside town. One day a young woman went to the well to fetch water, and the well heard her crying, and asked “What’s wrong?”
She stopped her sobbing and asked the well “You can talk?”
“Yes,” said the well. “Long ago, the witch who lives in this town gave me life so I could serve as a guardian to the townspeople.”
“Alas,” said the young woman. “I am the daughter of that witch. She lived in peace with the townsfolk for many years. But the new mayor, who is a violent and hateful man, riled the people up against her, and they burned her at the stake. I am young and still do not know very much magic. I tried to curse them, but my curses fizzled. Now I worry I will never avenge my mother’s death.”
“Do not be afraid,” said the well. “I will take care of this.”
The next morning, when the Mayor came to fetch water from the well, he heard an odd noise coming from the bottom. He peered over as far as he could to see what was happening. Then an impossibly long arm shot up from the bottom of the well, grabbed the mayor, and pulled him into the well shaft. There was a horrible crunching sound, and nobody ever saw the Mayor again. The townsfolk apologized to the witch’s daughter, and they all lived happily ever after.
Moral of the story: living well is the best revenge
I have a class of high school girls who will have a couple of free periods coming up after they're done with the finals in my class, so I offered to show them a couple of movies.
I probably shouldn't show them my favorites from the 80s. I also probably should show them my current favorites, many of which are rated R. And Deadpool is right out.
So I asked ChatGPT, "Recommend some movies from the last twenty years that are PG-13, 100 to 120 minutes long, and good for teenage girls."
It came back with a list of 10 movies.
All but two of them failed one of my criteria. They were either too long, too short, from too long ago, or Rated R.
80% failure rate.
I tried phrasing it a little differently, stressing the criteria again. It came back with a similar list, again with a high failure rate, but this time it pointed out "Doesn't meet your PG-13 requirement, but this is for reference" or "From 35 years ago, but this is for reference."
Maybe you could argue that my criteria were too strict, that there aren't 10 movies that meet them. Ok. Maybe. But then don't give me ten where I have go double-check them all because it's clear most of them shouldn't have been listed.
Maybe even say, "I could only find these three. If you broadened the date range or the length range I could find more." Because then I could say, "sure, surprise me", but I could also say, "No, these two look good, thanks."
I was listening to a podcast (Vergecast, FWIW), where one of the folks said that these generative AIs were like having a bad intern. You give them a task, and the work they do is so bad that you spend more time fixing it than if you'd just done it yourself from scratch. At least interns get better, but ChatGPT doesn't seem to be keeping pace with the intern curve.
This reminds me me of when we were designing our kitchen with a lot of blue in it. I said, "It's going to be either bold or garish, and we won't really know until we see it."
Hi, you probably won’t see this but I’m doing a debate at school for the most influential person and I’m going against you with Siddhartha Gautama, AKA Buddha. Would you say you’re more influential than Buddha?
Netflix/etc *do* have to pay for the extra bandwidth that the extra user is using to watch something you wouldn't have. So it *does* cost them something.
Mind you, it's not the amount they're quoting. What they're quoting is the lost income they would have made if all those users made their own accounts, but just because you reject that logic doesn't mean it costs them nothing.