Drew Ackerman of the "Sleep with Me" podcast talks with us about his new series "Dreaming of Bob Newhart" and how he's been using AI like ChatGPT to help with the creative aspects of his show.
Drew Ackerman of the "Sleep with Me" podcast talks with us about his new series "Dreaming of Bob Newhart" and how he's been using AI like ChatGPT to help with the creative aspects of his show.
COMPUTER-GENERATED TRANSCRIPT
MIGNON: Grammar Girl here. I'm Mignon Fogarty, and today I'm here with Drew Ackerman, who you probably know more as Scooter from the "Sleep With Me" podcast, one of the most popular and longest running sleep podcasts. Just a pleasure. Drew, welcome to the "Grammar Girl" podcast.
DREW: Thanks for having me on. I'm excited to talk.
MIGNON: You bet. First I have to thank you for helping me sleep on occasion over the years. Your show has been running since 2013, and you have produced I think more than 1,200 episodes. And I was thinking about what do I want to talk about, to Drew about, and I want to talk to you about your writing and your new project: the "Bob Newhart" recap with ChatGPT. But first I want to talk to you about the writing and, you are producing material that is at least in theory meant to be boring and, but I feel like there's maybe more to it than that too. So can you talk about just how you think about what you're putting together for sleep with me every episode?
DREW: Yeah. And you might be able to relate to this in a sense. It's a lot of "Sleep With Me" is content is designed, like thinking about the listener experience and what their experience is, and it has evolved over the years as I've made the show and gotten feedback and listened to the episodes.
But the whole, when I started "Sleep With Me," the idea was all these kids are getting bedtime stories, but there's nothing for adults. What's up with that? And not that kids don't like bedtime stories, but I always thought I think an adult bedtime story. Will be, it would be a bit different.
Like kids are more, they want to ask a lot of questions. They're easily distracted. So there were like even more highly engaged. And I was like, man, I think of an adult bedtime story, especially when you talk to people who can't sleep and you hear how they repurpose stuff like TV shows or comedy, that's what I used when I was a kid, not to be able to sleep is listening to comedy on the radio.
So it's okay, you're looking for something that's more friendly and distracting. Would people want to listen to that? And I guess as we've seen over the years of making podcasts for a long time, it's like everything is kind of niche, right? And so even as I started out, there wasn't this big boom of sleep podcasts or whatever, but now there's a lot of other sleep podcasts, but they serve different needs and different styles and different tastes.
And that kind of circles back to the writing in a sense of my show is capturing the feel of a friend you would call and be like, "Hey, tell me a story. I can't sleep. I'm not going to really pay attention to you, but I want some distraction." So get be boring. But not so boring. I know you're being boring, boring on purpose.
And that's why I love making the show year after year is that's a puzzle to solve every single episode.
MIGNON: So how do you do it? How do you walk that line between, just boring enough.
DREW: Some of it, unfortunately, probably some of it's maybe 50 percent natural talent. And then overpracticing the skill. So nowadays I'm more hyper aware when I'm at a social function, it's maybe I'll just do some listening because when I start talking, I may go on tangents and people like, excuse me, I got to use the restroom.
I'll be right. I'll be right back. Don't worry. But I think it's paying attention to details or specific things and delving into them, but also continually moving on. So I try to have structure for everything. Like some shows are more structured than others, but some kind of loose structure so that when it gets too boring, like I can almost develop a sense I can move on to the next topic or the next part of the story, but in between there, there's this freedom.
If a word pops in my head, like "malaise" that every six months that words pops in my head. And it's okay, and I'll start to just focus on the word and get in and I'm distracted, but I can lean into that distraction for long enough. And sometimes I push the boundaries and I've said, yeah, I just said "malaise" like about 86 times in the last eight minutes.
So some of you may be fed up, but it's another thing is being playful to take the seriousness out of bedtime.
MIGNON: So I'm guessing that this show isn't completely scripted, how do you put it together? Do you have an outline that helps you keep on track or is it just completely improvised with just working within your structure that you already have? How does it work?
DREW: So we over, when I first started making the show, I had no idea what I was doing and no idea what I was getting into and no experience. So it's "Hey, let me just keep making this and listening to the episodes." And as I started doing that, and I was working a full time job, I realized okay, at first I was going to write every episode as a story and then read it loosely.
And that didn't quite work because it felt a little bit too intentional. And then I realized I was putting out like multiple episodes a week. It's okay, writing every episode in a hard script, like even getting to a second or third draft is just not possible. And so now we make three general styles of shows, like shows that are scripted or soft scripted.
That I write and maybe they're only, they only get to a second draft. So I'm almost rewriting it as I'm telling the story, but those follow a general narrative. Then there's kind of rewatch podcasts where I'll watch something and take a lot of notes and be like, okay, what's interesting about this?
Like we did, we've done "Star Trek The Next Generation" for a long time. So it's okay, what are the performers doing? What's the set doing? What's happening in the story? And again, looking for those moments that might slip by because they're not super interesting, but it's like, "Hey, Jordy seems like he's in a bad mood. What's going on?"
We'll got to just think about that and go down that. Meander. And then I have improv podcasts where I'm sometimes it'll be a personal essay. Sometimes I'll find a random collection of stuff and kind of just go off tangents on that. Those ones are open and it uses three different skill sets, which I found alternating.
Let's me be productive, but it's also each listener has their own tastes. So some people prefer one style of show. So it also offers the listener reassurance Oh, if you didn't like this episode, a different style episode is going to come out soon. And it just worked out by accident but I'm grateful for it.
MIGNON: Yeah. That's one thing I've known. It seems like you have a really special close relationship with your listeners. You might have more back and forth with them than the average podcaster. And it seems like you have different ways of packaging your work that works differently for different people.
Can you talk about how you developed that really seems like a really deep understanding of your audience and what they need?
DREW: Yeah. I think that's a good question. I think part of it is coming from. I started the show because i couldn't sleep as a kid, and I felt really lonely and in pain, but I wanted something pleasant to distract me, and I also knew what it was like to tell people I couldn't sleep and have him not be able to relate to that they're like, "Hey man what did you think of something nice at bedtime and thinking about school?" And it was like, if I could do that I wouldn't be, I'd be asleep. And so like those kinds of things made me curious about the show. And I was also very critical. I was like, "Oh, this is a terrible idea." Like when I first thought of it, it was like, I'm not going to do that.
Even though, even in whatever, when I saw the idea, it was like, wait, two, 2011 it was like, "Oh, there's already podcasts about everything that you can enjoy." And that's what made me excited about it. But it was like, "Oh no." And I kept putting it off. But then when I started making it and even getting one or two pieces of feedback to know, "Oh wait, there is a couple other people out there looking for something like this."
So let me just keep trying. And I guess like a product based word is like iterate, but it is like a evolution of “Hey, let's try this next time. Or let's try this.” And that's I keep using the word puzzle, but it, for me, that's one of the childlike things I like about it's okay, this person really had a strong reaction to this.
So maybe it's like money or something like saying, okay, how can we change it in the next episode to still talk about. The same subject matter, but refine it or use a different word for it or anything like that. And so I think it is a constant returning. And usually for me, it's by making mistakes where I'm constantly returning to.
Putting myself in the listener's shoes and then trying to connect with them. I'm even going through that now where I was like, man, I think I lost perspective there for a while of like producing the show and making the show and trying to make it work. And then I forgot, wait, isn't my job to try to just connect with one listener a day or something like that?
And that's what I'm trying to return to. Cause it's that's really the entire relationship that makes it work in the end. People let us right up against their brain in their earbuds. And that's a pretty special relationship, whether it's during the day or at night.
MIGNON: Yeah, it's interesting that you brought up words because I was going to ask you about that. With, I don't know, I was thinking about like with Halloween coming up and I was like, I bet Drew can't do like scary stories, and I was wondering like what else do you feel like, and you mentioned like money can be something people might not want to be reminded of when they're trying to go to sleep.
Like what are some other things that maybe are some limitations on what you feel like you can talk about?
DREW: Halloween is a good one because it, and this is a good, like I'm working on an episode now. So my daughter, I have a daughter who's a teenager. And as you became like a guy like to bike around 15, she was like, I really going to haunted houses and in this current era, there's like really high quality haunted houses that you could go to.
So we started going to these haunted houses. That's like our fall hobby. And they were so detailed that I was like, man, this is so interesting. And there's so much work that goes into these. These are really make interesting episodes for the podcast, but the subject matter is not sleepy at all.
And and again, it was like something where it was just stuck in my brain. I'm like, man, I think I got to do an episode about this. How am I going to do it? So it is like rephrasing and reframing of instead of Halloween, I call it "autumnal joy." And it's instead of a haunted house, it's a place where they do an interpretation of autumnal joy.
And it could be a different take on that. I think there was one we went to that was called wax works and you could let your brain fill in what's going on in there, but I was like, Oh, this is how could wax be made into the shapes of autumnal joy? Or how's this going to be a celebration of all gourds and things like that.
So again, it becomes like a challenge in, and I don't know, this is just a speculation, but I think. That listeners want to be respected and not and there's something about towing that line that it's sometimes I may cross it and make a mistake in the listener was who will let me know, but it's not like I'm saying you can't handle this.
It's “Hey, let's reframe this.” You know what I'm talking about, but I'm going to make it in a way where I'm making it for you and I'm making these changes for you. Within the rules of our thing. But again, it's this is, and a lot of kids listen to the podcast now. So it's not like anything happens in the podcast that's adult in nature.
It's just geared towards adults,
MIGNON: yeah. Were you a creative writer before you started the podcast?
DREW: So like in school and stuff, I was like, I was very interested in writing and I, even as a kid, I had a big imagination and I was big into reading and I also had dyslexia. So it was like this disconnect with School. I love to read but I approached it differently. And I think in some sense that impacted the podcast in a positive way.
But I also, as I started my adulthood, I was a person that more thought about writing than I actually did any writing. And I struggled to follow through, but finally at some point I was following through more and more in writing in my spare time. And it’s not, and just like trying different things and trying more to be more silly.
And then this podcast actually developed out of some other thing I was writing that didn't work out where I was like, Oh, what am I going to do with my time now? And I was like, Oh, why don't you make that sleep podcast? You've been thinking about
MIGNON: Yeah. So the what are some of the, so we're going to talk about your "Bob Newhart" TV recap and you said you did "Star Trek Deep Space Nine."
DREW: We did TNG. This "Deep Space Nine" is not a lot of people's wishlist though. So
MIGNON: Yeah. So what are some of the other creative what are some of the most creative episodes that you feel like you've done?
DREW: I, I like let's see, like I always have, my favorites are always the ones that aren't the listeners favorites. Like it's funny. Because for me, it's for me, my favorite part about making it is when I sit down and make it and I'm surprised, so this is still the episode. Like it was a random kind of improv episode where I was like, what would happen?
And this was actually funny beause it was before there was AI was a thing. So it was like four or five years ago. I was like, what would happen if I interviewed a username generator, like something like just a piece of code that's job is to generate usernames. Cause I must, I'm like, "That must be the worst job on the internet because of all the people that think they're so funny, and they're putting in these user names that they think are so funny and unique."
And it's "Oh no, 4,000 people have thought of that username before you." And so I was like, let's sit down with a username generator and interview them. And then it was like, "Oh, what's your name?" "Oh, Jenny, the username generator." Okay. That makes sense. And it just I was like, I didn't know how it would go because I was imagining it, but it went differently and it was just like, I don't know where I was like, where's this even coming from?
Is this a collective unconscious or whatever? It was like this username generators answering questions in an unexpected way. So I really like that. But I'm, I've been, I'm always interested in what, what piques my curiosity when I'm out there, like just taking my dog for a walk or whatever.
And then usually it'd be like four months later, I'll be like, Oh wait, what am I going to do an episode about next week? Oh, okay. Wait a second. You were wondering about that. And it leads into that, but if some, I try to be open minded. So it's just about anything.
MIGNON: Yeah, I find myself wondering do you have to get in a mood before you start? Do you have a space or like a time of day? Do you always do this in the morning or at night? Like, how do you get in the meandering sleepy zone to make your show?
DREW: Yeah. So I do it in a specific space, which is like a client, like it's like under this staircase in my place. That's like not a full size staircase. It's three quarters. So when I get under there, I'm like, my shoulders are almost touching the walls and it's very quiet and perfect for making a sleep podcast.
And then I really hard for me to lower my anxiety. Like I tend to be a kind of hyper person. So when I get in there, sometimes I have to pause for a little while or do a little bit of prep. And remember why I'm there. So like to talk to one person and help them fall asleep. And so whatever I'm worried about or whatever I'm stressed about, or even though I've done this a long time, there's still some part of my brain.
That's you better make the best you to this episode better be perfect. You better make the best sleep podcast episode. And it's okay, that's not our job really. So like remembering those kinds of things. And it is like putting on clothing or something at this point where it's okay I need to go down there.
I got to put on the clothes Oh, yeah, I've gotten all those things and I got to do that and I got to look at this spreadsheet but okay, let's sit down there and let's just do our best. And that's a weird trick. I don't know if anybody's like listening or watching. It's like just doing that.
A lot of times I'm like, "Oh no, this episode's not ready. Oh, it's this is going to be awful. It's okay let's just go down and start recording the episode for 15 minutes. And if it's that awful, then we'll work on the rest of it tomorrow." And then just seeing it from there. So it's, I like to say it's most well-adjusted part of my life.
Unfortunately, it doesn't always translate. I have trouble carrying it to the rest of my things, but I do my best, but it's like there, I could be like radically kind with myself and all those things. But sometimes I need more practice during the daytime.
MIGNON: I was wondering that too if spending all those years writing surreal or, making the surreal absurdist, show has had any effect on your life outside of the podcast? Has it spilled over into your day to day life in any way?
DREW: I think so because I think like for a lot of us that explore things maybe one of the reasons we're drawn to it is like a sense of and maybe it's just a perceived sense of like difference from the world or negative uniqueness. And that's one of the positive sides of the internet I've found is like, you discover, you put yourself out there and yeah, there's people that respond strongly and negatively to what you do, but there's a lot of people that are like, “Oh, hey I'm just like that too.”
Or I worry about that too. Like I talked a while about it. Does anybody else rehearse a lot of stuff before they go and do it? Like in their regular life, in their head, when they're walking their dog or they're supposed to be doing some other work, it's okay, I gotta go to that appointment. How's this done?
What am I going to say? To fill in the time when I've got to talk to my dentist so many people were like, Hey, there's somebody who's there's even a show like that. And it cuts through that loneliness. I think that every person feels and that's I don't know if that answered the question, but It is reassuring.
MIGNON: Yeah. No, I, that is one of the, your show is about making people feel like they aren't alone, which is wonderful when they're in the deep dark night. You have "Dreaming of Bob Newhart," right? And you said one of the things that you're doing with it is you're using ChatGPT as I wrote it down.
I had to find it as a, I can't find it, but as a creative process. partner, essentially, to make the show. And, I would just love to hear why don't you start with how you came to the decision to get, to try something like that with the show?
DREW: Yeah. It bounces back to a few years ago. There's a couple of podcasters that make this podcast called neutrino watch, which I think they say is it's, it was before there was AI procedurally generated podcasts, they say and Jeff and Martin. And I was talking to Jeff because he makes another podcast called here.
"Here Be Monsters," and he was telling me about what they were doing, and I was like, "Oh man, that'd be so interesting to take all these transcripts from 'Sleep With Me' and load them into machine learning" was what it was called back then. And he was explaining to me like, he's "Oh, it's so simple. You just do these Python things."
And I was like, "Oh boy I don't know if I could handle that." But I was still curious. What would happen? How machines and language and creativity interact. It's very interesting even back then. And so it was something I was always curious about it. And then as AI became more user friendly and more powerful at first I just played around with it, like with some of the AI voices, just to see what it would sound like an image generation, just for fun, not relate to the podcast.
And then the next time I used it. Was when I was doing research for a Dungeons and Dragons style podcast and I needed to like I'm not like a expert in Dungeons and Dragons or anything, but there's a lot of research that goes into like certain numbers and characters and stuff like that. And I was like, huh, I wonder if AI can look this up quicker than something else.
Otherwise you have to use a reference book and it's got to go to the index, and if you got to find 300 things, it's like that's going to take a long time, and I was like "Hey, can you find these characters are these type of creatures with these kind of stats and it came back with the information?" and then I liked I'm like "This is actually correct," and so it and then I was like "Hey, can you do this?" Another thing with Dungeons and Dragons that I'm not particularly good at is math. So I was like, "Hey, can you run these math simulations with these rolls of random dice?" And it could do that a little bit imperfectly, but it still was enough that I could use it and then rewrite it for my purposes.
And then I had this idea for a podcast there's a show called "Newhart" that we're talking about in the "Bob Newhart" show that were both starred. The comedian Bob Newhart passed away this year. And "The Bob Newhart Show" preceded the "Newhart" show, but when the "Newhart" show ended, we had the final episode revealed that the entire show was a dream had by the character in the first sitcom and I was like, huh, that's interesting. And I thought about that as a rewatch podcast. I'm like, wait so they're saying retroactively every episode was a dream. What would a dream expert like think of this?
If they sat down and watched it and because I'm putting out multiple episodes a week, it's "Okay. I don't have the resources to sit down with a dream expert for two hours a week and do the," so I was like, “Huh, maybe I could use a ChatGPT a little bit to just do look up some of the information for me.”
And then I was like, wait a second. "Can you make," and I was learning as I was doing, I don't have a lot of AI background, but I was like, "Can you train this thing to pretend to be an expert, a bad expert in dream interpretation?" And I think the answer is but it's okay, I want you to know this about dreams and this about dreams, but also I want you to think about, I kept saying, I need out of the box ideas.
So I want you to think it like out of the box ideas. It'd give me random, strange things versus just straightforward, repetitive information. And then for me, that's like a raw material. So then it's okay, and the more randomness or strangeness, the better beause then I'll use that to go on more tangents.
But it's just also funny to be inputting this information is okay, I'm a Vermont innkeeper and I'm getting ready to celebrate Thanksgiving. It's just the more layers as I am describing it. The more opportunities for me to go off topic and get mixed up and for people to fall asleep,
MIGNON: Yeah. I'm just so curious about the, like the, so it's, did you have trouble getting it to be out of the box and outrageous? It sounded like you might've had to give it extra prompts to be like extra weird or.
DREW: Yeah, I find that it's, it can be really weird sometimes. And then other times it can give you a lot of the same repetitive information or forget. It's like a very imperfect coworker. And it's also like, where are you getting this information from? They can never answer those questions like to my I'm like, Oh, it's based on Publicly available research, these books like but it so sometimes it is, I think over the last six months, even it used to be that it wasn't a time saving thing, it was like, Oh, this takes me as much time as it does to do it another way, but now it's it is becoming more efficient.
But it does take some prodding to be like, Hey, can you give me some out of the box ideas? No, not the same ones that you did those last time. Like it became obsessed with like alternative universe theory. And it was constantly like, Oh, like this dream could be like the, you're reflecting on your existence in another.
And it's okay, but you said that already five times. So give me something a little bit different than that theory. Also a language thing, I think is like. I noticed myself using please and thank you it, I just, and then once I was aware of it, I was like, I better not stop because if this thing becomes sentient down the road, I wanted to remember I was the one that said, please.
And thank you. And when it comes to do us all in and be like, can you just be a little bit more gentle with me?
MIGNON: Yeah, I found that too. So are you running it all? So do you have one long chat going where all your stuff is together? So it has the memory of what you've been doing. Are you doing like multiple different chats for each like episode or even more fragmented than that?
DREW: Both because I found that when I was telling it to remember, sometimes it would get mixed up and I guess those are hallucinations maybe, and it would be talking about other things. And then I'll be like, hey, if we start a new chat, can you remember what we talked about in the other chat?
It. This is like a coworker's totally don't worry at all. And it's okay, I better worry then I better have those prompts saved to another service because and so I've, I'm still experimenting and I'm still working on getting it to replicate dreams. It, it straight up says oh, I cannot dream.
Like I can only simulate dreaming, and then I'll try to get it or to try to. It is fun because it's you can't force it to hallucinate or be weird. You have to work with it in a gentle way. But when it goes off topic, like it, it gave me one episode it with total, that's the other thing that I think as a broader issue is Concerning is or maybe it's just my projection.
It gives answers with total confidence that are not based in reality whatsoever. Like I said, Hey, can you summarize what happened in this episode? We're talking about it. It gave me a summary of an episode that it was like, this does not exist on planet earth, at least.
MIGNON: Yeah, no, it's interesting that you're using it, in this creative way because it's known for hallucinating and do you find, did you find that you were like starting to trust it? And then when it came up with a summary that was completely wrong, were you shocked or did you remind yourself like, “Oh, I remember now it's not great.” Things like that always. So I was wondering how you're thinking about it that way.
DREW: I think when I realized it was more of a disappointment for like future work that it could have saved me. It's Oh, if you could give me summaries about all this kind of stuff officially, that would save my time. But it's okay, you're obviously can't be trusted with this type of work. Cause I tried it another, I was doing a recap of the 2012 Macy's Thanksgiving day parade.
And I was like, Hey, can you give me a list of all the floats? And performances. Are you familiar with this parade? Oh yeah. I'm real familiar with it. Okay. What's your source? Oh, generally available information or whatever. Okay. Give me a list of everything. And it gave this list and I was like, this is, it had like timestamps and everything.
I'm like, this is not the correct parade. So it's pretty funny, but it's also Okay. I feel bad for students trying to get around, to misuse it because, and for teachers, because it's hey, I'm sorry, but this is just not this is not great expectations. This sounds more like a twilight, like you're writing about even though you have it all quoted and everything.
MIGNON: Yeah, no, it is. It is disturbing how confident it is. It's like it really, you have to be on your guard, you said that it's gotten more efficient. Your process has gotten more efficient since you've been using it. Do you think is that because the tools have gotten better because they're evolving so quickly?
Or is that you feel like you've gotten more efficient at using it yourself?
DREW: I guess I've gotten, probably gotten more efficient using it myself. Maybe it's gotten more efficient behind the scenes, but it's I've more, I guess it's another type of language learning. It's Oh, what is the language this wants from me? That's going to help it. And even when you, it's a good footy, it's “Hey, could you resummarize what you just did in a prompt?”
And then it gives you like two sentences and then it's okay, I want you to replicate something similar. Is that going to be enough? Totally be enough. Don't worry about it. I got it covered. So it's I think I'm getting better at learning what it likes to hear and what it doesn't, and then how I can get it, because even when it says like memory updated, like that's another one.
It's don't worry, boss. Got it right up here. I got it covered. It's okay, remember I told you, like I tried and wanted to try it to do some astrological stuff as a part of the dreams. And I was like, okay, this is when the show started. This was the date of the pilot. So let's pretend that's the birth date and use that for the astrology.
And then it's, that's one of the things I keep having to return to. Remember, you're going to base it on these dates or the Ngram. I said, "Hey, like I took some Ngram tests, imagining I was Bob Newhart". And I said, "Okay, this looks like my Ngram. Can you use this at all in the dreams?" But then it gets mixed up and it's "Oh, this character probably represents this," it still understands it, but it just latches onto the wrong thing sometimes.
MIGNON: Are you using custom GPTs? When you said memory updated, it made me think of that.
DREW: I don't know like I'm just using the main the one with this circle type thing and I'll start a new chat sometimes or and yeah, it's every once in a while, I'm like, Oh, let me watch some YouTube videos or something like that. But then my part of my brain's like. You got to get back focus on the work, man.
Come on. This is so yeah, it is an interesting use case for me because I'm not the most efficient person at using it either. But in a weird sense, that kind of serves my purposes at the, like the fact that I'm bad at it also helps it in just this use case only.
MIGNON: Yeah. No, the thing about the custom gpt is it's like you could preload information into it and it's supposed to remember what you tell it, and then you can go back to it and you don't have to use such complicated prompts. But I, and it would say memory updated when you add things and, but it didn't work well for me.
I kept telling there was this one thing I was trying to do and I kept saying, keep it to 250 words or less, and it would be like, memory updated, and then it would gimme like five paragraphs. And I'm like what are we talking about? I just, you just said, and I'm essentially stopped using those because they worked so poorly for me. But it's working really well for you in this particular way do you think that you'll use it in other ways? Do you imagine that you're going to work at use it in other ways more in the future?
DREW: Yeah. I think for me the next way I'm going to use it is for randomness, like visual randomness to, for me to turn into audio. Like when I first played around with it, like I thought about okay. We don't know necessarily where these art prompts are getting all of their art from, but one place I'm comfortable with is like commercial advertisements.
So I've been playing around with it of "Hey, are you familiar with these kinds of advertisements? Toy advertisements from the 1980s." Oh yeah. "Okay. Let's build some prompts based on toy advertisements from the 1980s" and then inputting like stuff from "Sleep With Me," like an old "Sleep With Me" transcript or something.
And then seeing the result, And those results are so strange. And if there is an alternative universe based on ChatGPT's dreams or whatever, I do not want to be there. But it does give me an opportunity that it's developing these bizarre images based on something. We're all from like archetypal like a toy catalog or whatever.
That I think it offers an interesting opportunity for me to be like, dig into the details. Even if it's not related to the ChatGPT and just look at the image and say. Okay. What's happening here. What kind of tangents can I go on? Or who knows down, cause it's just like, it's just really strange.
Like it became obsessed with pretzels. So it just kept giving me these images of toys playing with pretzels and surrounded by pretzels and sitting on pretzels and I was just like okay. Interesting.
MIGNON: That's wild. Do you to finish up, do you have any advice for writers who might be thinking about experimenting with ChatGPT as like a creative partner in their writing?
DREW: Yeah. I would say. Be cautious, there's opportunity there, I think and unfortunately, I think all of us that were presented with the already open Pandora's box, right? It's Hey, this box is open we're going to ask for permission later or whatever or figure it out. Yeah. It is a tough spot to be in, but it is something that's it does have, it is a new tool and maybe it's a tool we have mixed feelings about, or maybe it's a tool that feels, even for me, it's will this thing start developing podcasts that could put people to sleep?
What would that be like? I've done those experiments and it's on, that was a couple of years ago, but it's there's that too uncanny valley. So I think there are tools. Research, as long as you're double checking, the research is good or randomness or even for like basic like a lot of times you're trying to do Brainstorming like, "Hey, I want to come up with a thousand names of people and build off of that or a thousand places" or, "Hey, what are places that have this climate?" or things like that.
I think it is way more, and that's what ended up with the D and even the dreams. It was like. This is so much easier to use than a search engine for those research purposes, and I was a librarian for a while, so I mean my old coworkers are probably … but it's like it's just a reality is like a like I would love to spend eight hours a day in the library doing work like that and reading dream interpretation things.
But it's "Hey, I'm not presenting this as an actual in depth analysis of dreams." It's more of like silly tangents about this. And it's so how can I get the information to do these silly tangents? And it does work well for that,
MIGNON: Wonderful. Yeah. And then to finish up, we usually ask our guests if they have any book recommendations. So are there any books you just absolutely love that you'd like to tell us about?
DREW: I would say books. I absolutely love all the books I'm reading recently are not sleepy at all. Like I've been reading like Philip Roth. I never read. Before which I really enjoy, but I don't know if that helped people at bedtime right now. The book that changed my life it was in, it's not considered one of his best works is "Hocus Pocus" by Kurt Vonnegut.
Like I just read it at this time in my life when I was like in early high school and I didn't really. I was only familiar with the literature as presented to, middle school students. And I read it and it's like a fake out autobiography written on scraps of paper. And I just had never been exposed to something like that.
And it opened up my mind in some way of wait, you can, authors make fictional biographies where the, the character is telling their biography or writing their autobiography and that's the story in that it was like written in this strange segments and stuff like that, and that it could tell.
Even though it's fiction, that there was a lot of research that went into it like about the sense of place and the sense of the industries and the sense of the changing of the places cause it took place in upstate New York in the Finger Lakes area. And yeah, that's a book.
I read it a while ago. I think I reread it, read it every couple of years, but it was one that just changed my life. Just by exposing me to something different than what I'd been exposed to before.
MIGNON: Super. Thank you. Thank you for, sharing your story with us today and your experiences with the ChatGPT experiment. That's really interesting. Where can, where do you want people to go to find out more about what you're doing?
DREW: Yeah. If wherever you're consuming this podcast, you could search for "Sleep With Me" and check it out. And also I don't always share this, but yes, like sleep, "Sleep With Me" is like a podcast to put you to sleep, but it may not work for you. So even in my podcast, I have a website listed with all the other sleep podcasts out there so that you could find one that'll help you out.
But Yeah, check it out in your podcast app.
MIGNON: Great. Thank you. Drew Ackerman from the "Sleep With Me" podcast. Thanks so much.
DREW: Thanks for having me on.