Is AI a useful tool? Is it something we should be worried about? Will it replace us?

These are all questions it’s important to be asking nowadays, but we can’t ask the internet because, frankly, its answers will be suspect.

I think that’s why I’ve gotten a lot of requests in the past year to write about it. I don’t want to, because I’ve only tried using it about three times.

The first time was a little over a year ago, when I was writing an article about painting a wall. And I kept hearing that AI can write whatever, and even imitate the style of a specific writer.

So I typed, “Please write a funny article of 1,500 words in the style of Mordechai Schmutter about how to paint a wall.”

So first of all, what it produced was not 1,500 words. I’m not as good at math as computers are, but I say it was 400.

So I wrote, “That wasn’t 1,500 words,” and it said, “Oh!” and it produced a second version that was 700 words

If I wanted to argue about word count, I would ask my students to write my articles.

In fact, my students are always asking me, “Why do I have to learn to write? Can’t I just use AI? If my essay was AI, how would you know?”

Well, if it’s dead boring but has perfect grammar, then I’ll know. No student who’s that good at half of what I teach is that bad at the other half.

I had to read through several versions of the article by AI, and it was bad writing. What I write is bad writing too, but at least I disguise it with jokes. AI doesn’t know what jokes are. There weren’t any jokes. It was an upbeat article about how to paint a wall, and then it said, “And now you know how to paint a wall the Mordechai Schmutter way!” And I said, “That’s just… the regular way.”

Also, one version, for example, contained a story about how my little dog walked across the paint and left footprints everywhere. I don’t have a little dog. I can’t mention a dog once and then never mention it again.

But how’s it supposed to know everything you need it to? It doesn’t know all your stories. Basically, what I found was that if you want to write something using AI, you have to put in so much information and so clearly that you’re basically writing the piece yourself. You’re just writing it out of order. AI is putting it in order, and also inserting a story about a dog because you forgot to tell it—in your prompt that was longer than the piece of writing AI spit out—“And I don’t have a dog.”

So I said, “This piece is not in the style of Mordechai Schmutter.”

And it said, “Mordechai Schmutter is a fictional character I made up for the purposes of this article.”

And I said, “I gave you that name.”

So it said, “Then you made him up.”

And I said, “I am Mordechai Schmutter.”

So I had to have this whole fight about whether I exist in the real world with a computer program that doesn’t exist in the real world.

But it does learn. I recently asked about myself again, to see how I’m doing, and apparently at some point someone taught it to say the following:

“Mordechai Schmutter is an Orthodox Jewish humor writer known for his witty and humorous observations on family life, Jewish traditions and everyday experiences. He has written for various publications, including Ami Magazine, and has published several books.”

Which… Okay, they’re getting there. I’ve never written for Ami. There was someone who wrote for Ami who in one article several years ago claimed to be Mordechai Schmutter, as a joke. That’s the guy the AI picked up on.

My second attempt at using AI was for my Presidents’ Day column. Every year, I write a “fun facts” article about a given president, and last year I was writing about James Monroe. So I asked it, “Please tell me a true, funny story about James Monroe.” And it gave me a story about how Monroe was once at a formal dinner in France, and he was served an exotic bird and he choked on the bone, and the bone went flying into the plate of a foreign dignitary, and Monroe quipped, “Monsieur, it appears your bird is not only difficult to swallow, but also difficult to digest!”

Get it? Me neither. But at least it’s a story. And I said, “Wow! I couldn’t find this story anywhere else! It’s a good thing I used AI!” And then I said, “Wait. I couldn’t find this story anywhere else.”

Only AI knew this story from 200 years ago.

So I asked, “Is this story true?” and it said, “It’s often recounted in historical literature about Monroe, but like many tales from the past, its accuracy may be uncertain.”

So I then asked, “What’s your source?” and it said, “The story is a common anecdote found in various historical accounts. It’s often cited in discussions about Monroe’s life and presidency.”

So I wrote, “Did you make it up?” and it said, “I apologize for the confusion. Yes, the story presented was a fictional creation intended to illustrate the kind of light-hearted anecdotes that might be associated with historical figures like Monroe.”

Apparently, when I said “give me a true story” and it gave me a story and I asked, “Is this true?” and it said yes and I asked for sources and it said there are many sources and it’s often cited, I was just confused.

So I said, “I told you to find a true story,” and it said, “There are no funny true stories about James Monroe.” So I said, “You could have said that,” and it said, “I apologize for the confusion.” Which is not an apology.

So it doesn’t understand certain words, like confusion. Or true. Or apologize.

But then how do I ever know it’s not making up answers? Like if I ask it how to safely change an electrical socket, how do I know this is the real way, and it’s not just guessing?

So I gave up asking it for help with articles, around the time some rabbanim were warning us against AI, possibly because of similar experiences. Or they didn’t want people asking it halacha shaylos. It’ll make up halachos, and then when you go to Gehenom, it’s gonna say, “Oh, sorry for the confusion.”

(It’ll be there too, obviously.)

And then I heard there was a kosher AI. So I decided to try that out. And what am I going to try with a kosher AI? I asked for a recipe.

I wrote, “Please give me a moist, zero-carb chicken cutlet recipe that I can make ahead of time for a Shabbos night seudah.” And it gave me a recipe that included parmesan.

Should I have specified that I wanted the Shabbos seudah to be kosher?

So I wrote, “I can’t put parmesan in a chicken recipe.” And it said, “Oops! Here’s another recipe.” And then it gave me the exact same recipe, but without parmesan.

Wow, thank you. A fleishig recipe is not just a milchig recipe without cheese. You can’t say “Fleishig lasagna recipe: Noodles, sauce, noodles, sauce, noodles, sauce.” That’s not lasagna.

So is AI like people? Let’s see: It has to be told a million times what to do. It lies. It resorts to things like, “They say” when it makes up a story. It doesn’t know how to apologize. It’s easily confused and accuses the other party of being confused. It doesn’t believe Mordechai Schmutter is a real person.

So basically, it is like people, but the worst people. Not the people you want to be around.

You don’t have to hire those people; you can use AI!

Sure, it can improve to eventually be the second-to-worst type of person, and so on. We should keep an eye on it. So I’ll probably revisit the topic at some point, unless AI has come for my job by then.

Good luck. It doesn’t pay a whole lot.

Mordechai Schmutter is a human Orthodox Jewish humor writer known for his witty and humorous observations on family life, Jewish traditions and everyday experiences. He has written for various publications, including Ami Magazine, and has published several books.

Mordechai Schmutter is a freelance writer and a humor columnist for Hamodia and other magazines. He has also published eight books and does stand-up comedy. You can contact him at MSchmutter@gmail.com.