Episode Transcript
Don (00:15)
All right, episode 138, we're back, Rob.
Rob (00:24)
Episode 138. We're back! It seems like just yesterday it was episode 137.
Don (00:29)
Do you remember when we used to do this every day?
Rob (00:33)
It wasn't a long run.
Don (00:35)
When we started the podcast, we rocked one of these things every day. I don't know either, but it lasted longer than you think it did, I think. That's what she said. We had a lot of things to say to a lot of different people, and now we don't have anything to say anymore. So we scaled back to, yeah, I think we recorded one of these about a week, I think. One or two weeks.
Rob (00:38)
Yeah, I don't know how long that lasted.
Yeah, well is this our first two parter?
Don (01:03)
I mean, we've had follow ups. I'd say this is a direct two-parter, ⁓ tailing on the excitement of yesterday's ⁓ AI draft. So we've had an update. So it literally is. It literally is the next day. And we have a follow up. We So I think you should just jump
Rob (01:07)
first half second half. Drama. It was a cliffhanger last episode cliffhanger and we are here to as a follow-up. I'm gonna start by reading an email
Don (01:33)
OK, here we go. This is exciting. we putting people on blast? mean, are we putting that out into the interwebs? Or no, we're still keeping it confident. OK, I'm in.
Rob (01:39)
There's no other way. I shall remain nameless. Okay, got it. I'm not putting people on blast. Yeah. Here's the email. Dear Mr. Broadfoot. Sounds formal. I would like to address the situation regarding the use of AI detectors and version history. While these tools can be helpful, they are not infallible and I must rely on the resources at my disposal. Ultimately, it was a judgment call and I made a bad call. ⁓ Bad judgment. I have had a conversation with
Your daughter and I firmly believe in her sincerity. Okay, I deeply regret the misunderstanding as her expression conveyed genuine emotion I will be retracting any documentation related to this matter And will extend my apologies to your daughter. Got it. I will prioritize resolving this issue promptly Should you have any additional questions or concerns? Please do not hesitate to reach out best regards so Quick recap. Yeah
Don (02:38)
For those that listen to-
Rob (02:39)
Five. There was an AI detection tool used against a, ⁓ an in-class essay that my daughter wrote. She's a sophomore in high school. Right. ⁓ and the teacher, ⁓ used the tool and determined that AI was in fact used to generate part of the paper. And she was accused of that plagiarism modern day plagiarism. call it. ⁓ to which
You know, my daughter vehemently denied this and was very, upset about it. As was I. It's been a stressed out 24 hours. When your child's accused of something like this and then you anyway, um, kudos to the school and the teacher for and could, well first kudos to my daughter. She stood up for herself, went and spoke with the teacher. It's a great lesson for her. Um, you leave in yourself, stand up for yourself. Um, and then kudos to the school for following up with.
Don (03:11)
There's been a lot of
Rob (03:34)
the proper response and resolution to the issue. So it's a happy ending. But part of the reason we were talking about this in the first place is just sort of AI, you know, sort of in that sandbox. And what I find interesting is this. So I ended up taking the portion of the essay that was in question, and I actually put it into chat GPT and said, hey, check this for grammar.
Don (04:04)
Okay, check it for grammar. That's interesting. Okay.
Rob (04:06)
Let's see what happens because if another AI, if AI did generate it, in theory, it should use proper grammar. And this is not a knock on my daughter's writing skills. However, the results were, there are four grammatical errors. And you and I discussed these for a moment after the podcast yesterday, but there were four grammatical errors.
Don (04:13)
It should use proper grammar. Yeah, the two things we discussed after the podcast was once the shock had worn off of like what's happening right now, right? Was hey, if you actually read this it it really doesn't feel like AI wrote it because Because of the grammar correct not that it's incorrect grammar. It's that it felt colloquial to a human being if I
Rob (04:50)
Yeah, it absolutely did. so, you know, as I was getting amped up today to go back to the school and, you know, plead my case and do all the things in support of my daughter, one of my key pieces of evidence was going to be, well, my robot said that there are errors here. ⁓ How can that be if a robot generated it in the first place? So we're going to go, my robot says this, your robot says this. Point being, the robots aren't perfect.
Don (05:20)
No, no, think I think for me the bigger ⁓ umbrella over this entire thing especially utilization of AI right is Is what I wrote down taking notes here was is the cons is not a concept but is the aspect of critical thinking Sure critical thinking right and we are in such an immediacy as a society. I do this and that thing I'm gonna
Deadline clients, right and it's just copy and paste of this write an email, you know, it's like if you know stop for one second Yeah, literally five seconds and just do like arguably the least amount of critical thinking possible and Apply it to the subject matter at hand and then determine. Hey, what are the proper next steps there? Right? So like to spin it around a real world here one of the things
that the youngest of designers that we hire from time to time or junior graphic designers or whatever like that, an aspect of critical thinking is, client came in and told you to make these changes. ⁓ hey, remove this paragraph. Or we've got this brochure or change this, this, or that, or whatever that. OK, well, let us just delete that paragraph and then leave that big honking empty white space on the right.
There is a trickle down factor of, hey, make this one change. But then you still need to stop and look at it and go, hey, wait a minute. Does this make sense? Exactly. Critical thinking. Does this make sense? Hey, AI said that these three sentences kind of maybe, sort of might have been done by AI. Well, stop for a minute and go, hey, does this make sense? And we're in this weird like, people don't have critical thinking anymore.
Rob (06:50)
This makes sense. Let me read this and think about.
Don (07:13)
They just immediately react. It's just, somebody said something. This is the way it is. And it's like, well, hold on a second here. Maybe that's not the best use of judgment. Or hey, let me just play devil's advocate for one second and say, hey, I know you asked me to do this, but what about this instead? Or things like that. ⁓ The robots, it's like we can't seed all brain power.
Rob (07:35)
robots. And I would venture to guess in this specific instance, if you were to write, Don writes a, writes an essay. yeah, it was written. Yeah. There are details that I've left out. This was written in class being monitored. on. Don you, you write an essay or a poem. Let's say you write a beautiful love poem wife. Okay. I guarantee you, if I took that and copied it, ran it through,
Don (07:43)
By the way, in class. Okay, so Don writes an essay. Hi, Co-
Rob (08:04)
Zero GPT or whatever and said did AI write this it would spit out There is a chance that yeah AI could have written this sure sure I doubt There's a very limited percentage of time where it's like absolutely not. Yeah, no way. Yeah, like it doesn't make sense. There's always going to be some Chance that AI could have written it I think I'm making assumptions
Don (08:27)
Yeah, mean, again,
I'm smart enough to know how it even works in the first place. is it, you know, is it checking pattern? Do humans write in patterns? And that's why yesterday you were saying, hey, the longer the text, the longer, the more runway we have, the easier it is for the machines to determine if it was created by a machine or not. But that is because humans write to a pattern.
Rob (08:51)
Well, I mean, let's find out. There's a reason that, know, Charles Bukowski sounds like Charles B. He's got a very unique, you know, it's why authors write the way that Stephen King, he's got a very unique style of writing. then, yes, so the longer the text, the easier it is for the robots to go, oh, I see the patterns, I see the language, I see the almost mathematics behind, in a way, mathematics behind the language and the patterns. And they can then identify things that ⁓ fall outside of that typical pattern Yeah, which is why one of the main points when I was researching this yesterday Was the very first point that came up, you know when you when you when you dive into what's wrong with You know AI detectors sure the first problem is well, it doesn't work for short. Yeah You know passages. Yeah, because they can't it's not long enough to identify a pattern sure. therefore it's gonna say well, yeah Yeah, certainly could have been written by a computer. I don't know.
Don (09:49)
I'm sitting here juggling because we, I'm shifting gears for one quick second. There used to be a Post-It note on that, in that desk. Do you remember what that Post-It note used to say?
Rob (10:03)
I who I posted it for, I'm not going to say that. No, no, But I do remember... ⁓
Don (10:08)
It said, write like a human. Yeah, right. Write like a human. And it was direction for a copywriter. That's direction. Which is so funny to think about only because it had nothing to do with AI. It was like, don't be mechanical about headline creation. That's right. It's advertising. There's a conversational marketing aspect about it. So it was like, hey, the feedback you gave him was write like a human.
Rob (10:10)
like a human. for a copywriter.
Don (10:37)
wrote that down on a Post-It note and put it in his office as a reminder to like, hey, get the facts out. You always got to your first draft. Get it out, get it out, get it out, get it out, right? And then it's like, hey, how would you just say this to someone instead of writing it as a white paper for a minute?
Rob (10:53)
And that's a great lesson for all junior writers because we have a tendency to when we first starting like, okay, this has to sound all professional and sound.
Don (11:05)
I know, I know. So random weird side note I thought about as that person's office. It does, it does. that was all, man, that was pre AI. I can't believe I just remembered that. Well, it was up there forever. And it's not, it's not up there right now. Yeah, we should have.
Rob (11:11)
Wow, We should have put it on the fridge. Lasting advice, yep, right? Like a human.
Don (11:29)
Right, a human. ⁓ critical thinking. don't know, are there any other aspects to critical thinking? think that, you know, ⁓ if we pivot into like advertising and design and...
Rob (11:38)
No, I think it's just that and you touched on it. I'll reiterate. ⁓ You we always tell our folks, you got to stop and think. You just got to stop and think. Yeah. And there's sometimes where, okay, know, whether it's, you know, if you're designing something, right, if you're creating an image, okay, well, does this, does it make sense? Does this make sense what we're doing? We're not, yeah, the client wants us to do this, but we can push back and we should push back.
Don (11:50)
Always just a reaction. And clients ask for the weirdest things on earth. They're like, I to mix these three things together and like, those don't go together.
Rob (12:13)
There's an example of when clients don't critically think, ⁓ Hey, here's a, ⁓ I don't know, 16 page word doc. Okay. And I need, I would like for this to a four pager, a four page brochure. And you just kind of go, huh, huh. Are we staring at the same 16 full pages of, of
Don (12:25)
This text like single space single barely any paragraphs
Rob (12:41)
They're also on 8.5 by 11 pages. So it's one to one what we're doing. And somehow we're supposed to fit that 16 pages into four pages.
Don (12:50)
That happens more than I'd like to admit. Here's a six page word doc and it's a front and back five by seven post. I was like, wait a minute.
Rob (12:54)
It does, it's crazy. What are we talking about?
Don (13:06)
Yeah, I think, you know, I don't know. It's an unfortunate consequence of just how fast paced everything is that, you know, we're just shuffling things back and forth and back and forth and back and forth to get to the deadline. We've got to the trade show, the this, the that, the insertion, the, you know, and it's like, whoa, whoa, whoa, whoa, on a second. Yeah, yeah. Stop and think for 30 seconds. Yeah, and that applies to the robots as well.
Rob (13:13)
It's Gotta stop and Does this make sense? So,
Don (13:35)
Anyway,
Rob (13:38)
Thank God the story ended this way. I was going to be really well, we wouldn't have published the podcast if it.
Don (13:43)
No, we would have just delayed our recording until there was.
Rob (13:47)
If it had gone the other way.
Don (13:49)
Yeah, yeah. We're in about, we'd be asking the public for help on what other AI sniffers to find out.
Rob (13:56)
Exactly. anyway, that's it. All right.
Don (14:01)
That's well all the robots can find us at mocktheagency.com. That's right
Rob (14:05)
And so can the humans. Just think. Use your brain and I bet you can find us.
Don (14:09)
type into the magical Google.
Rob (14:12)
We're going to leave it at that. Yeah. Trust the humans.
Don (14:14)
All right. Thanks, everybody.
