Experiments with Midjourney and ChatGPT

Right this moment on the Self-Publishing Information podcast: Experiments with Midjourney and ChatGPT. ALLi Information Editor Dan Holloway and Information and Podcast Producer Howard Lovy focus on the newest developments in synthetic intelligence. Apple turns to AI for audiobook narration, and many authors are up in arms over a youngsters’s e-book written and illustrated by AI. Howard talks about his experiments with AI in fiction and how the expertise could be included into your inventive work with out changing it.

Discover extra writer recommendation, suggestions and instruments at our Self-publishing Writer Recommendation Heart, with an enormous archive of practically 2,000 weblog posts, and a helpful search field to seek out key data on the subject you want.

And, in case you haven’t already, we invite you to be part of our group and develop into a self-publishing ally. You are able to do that at allianceindependentauthors.org.

Take heed to Self-Publishing Information: Experiments with Midjourney and ChatGPT

Experiments with #Midjourney and #ChatGPT: Does AI enhance or replace creativity? Self-Publishing News #Podcast with @agnieszkasshoes and @howard_lovy Click To Tweet

Do not Miss an #AskALLi Broadcast

Subscribe to our Ask ALLi podcast on iTunes, Stitcher, Participant.FM, Overcast, Pocket Casts, or Spotify.

              

OR, sign up to get notified via email proper when a broadcast is about to go reside on Facebook and when a brand new podcast is printed (#AskALLi recommendation on Fridays and indie inspiration on Sundays).

Present Notes

AI for Authors: Sensible and Moral Pointers

In regards to the Hosts

Dan Holloway is a novelist, poet and spoken phrase artist. He’s the MC of the efficiency arts present The New Libertines. He competed on the Nationwide Poetry Slam remaining on the Royal Albert Corridor. His newest assortment, The Transparency of Sutures, is available on Kindle.

Howard Lovy has been a journalist for greater than 30 years, and has spent the final eight years amplifying the voices of impartial publishers and authors. He works with authors as a e-book editor to organize their work to be printed. Howard can also be a contract author specializing in Jewish points whose work seems recurrently in Publishers Weekly, the Jewish Every day Ahead, and Longreads. Discover Howard at howardlovy.com, LinkedIn and Twitter.

Learn the Transcripts: Experiments with Midjourney and ChatGPT

Howard Lovy: Hi there, and welcome to the January 2023 version of Self-Publishing Information from the Alliance of Impartial Authors. I am Howard Lovy, ALLi’s information and podcast producer and e-book editor at howardlovy.com. Becoming a member of me is ALLi Information editor Dan Holloway. Hi there Dan, how are you?

Dan Holloway: Hello Howard. Joyful New 12 months.

Howard Lovy: Joyful New 12 months to you too. With 2023, that seems like such a futuristic date, we’ve the long run to speak about in at the moment’s podcast.

Dan Holloway: Sure. So, it is type of a particular version. We’re all tech this week.

Howard Lovy: Proper, all tech, on a regular basis. All people’s speaking about synthetic intelligence and whether or not it is good or dangerous, or a mixture of the 2.

So, a few issues are converging in my life. I made a decision to put in writing a bit of fiction for the primary time after spending my complete profession writing non-fiction, and on the similar time, I am studying extra about AI as a part of the inventive course of. So, I believed I would use AI in my work and see what all of the fuss is about.

Apple Turning to AI Narration for Audiobooks

However first, earlier than we go into that, let’s speak in regards to the information and what’s occurring on this planet of AI. I assume we will begin with Apple. They just lately introduced that they’re turning to AI for narration in audiobooks.

Dan Holloway: Sure. So, they’ve introduced they are going to have a particular assortment of AI voice-generated audiobooks. They don’t seem to be the primary to be engaged on this, it is the primary type of huge announcement of an precise assortment. So, Google have been working laborious on this. On the Self-Publishing Method convention final summer time, I spoke to the staff from Google and they’re working very laborious on this; they clearly see it as a part of the long run.

Clearly, the larger piece on this space is Spotify. They’ve purchased a text-to-speech era firm, Semantic. They have not publicly launched no matter they are going to be doing with it, but it surely’s clearly going to be one thing of the identical.

That is a kind of areas, I bear in mind simply over a 12 months in the past at FutureBook, there was a presentation explaining that it is a huge hole available in the market, that 95% of eBooks do not have an audio equal, and I feel this was SpeechKey who had been making this pitch. They noticed that because the hole that AI may fill as a method of bringing these books to the general public, and additionally of constructing narration cheaper for indie authors.

It is mainly takes a zero off the value of manufacturing an audiobook, although clearly voice narrators are going to have considerations. For a short while, I feel individuals had been ignoring that a part of the story, it felt like a bit little bit of a scarcity of solidarity. There are lots of people who’ve immediately obtained on this aspect now that authors are being affected. Extra had been getting concerned when cowl artists had been being affected, perhaps fewer than may need been concerned when it was simply voice narrators.

So, I feel one factor that we have discovered from this complete factor is that we’re a inventive group, and it’s going to have an effect on us all. So, all of us should be considering collectively about what our response to it’s going to be.

Howard Lovy: In the meantime, I used to be studying your column earlier this week, or final week, you talked in regards to the reporting of this story, and one reporter mentioned that the AI-generated narration won’t ever seize the magic of the human voice.

And also you say, properly, that is all properly and good till it does. There goes that argument.

The Limitations of AI Artwork

Dan Holloway: It was actually fascinating watching individuals’s response to AI artwork final 12 months. If 2021 was about narration, 2022 was about artwork, and this 12 months’s clearly about textual content. AI-generated artwork went from actually garbage, type of just a few months in the past, to being immediately actually good, which is, I assume, one thing you possibly can touch upon, as a result of I’ve seen your experiments with Midjourney, and the outcomes are actually fairly startling.

Howard Lovy: Yeah, they’re, they’re unimaginable. They’re additionally restricted, and it could possibly be as a result of, you already know, rubbish in, rubbish out, and perhaps I nonetheless must learn to do the right prompts, however a part of what I am doing is utilizing Midjourney to image what these characters I am creating would possibly appear like.

A number of the limitations that I’ve seen, properly, it may possibly’t do one thing sophisticated like one character taking part in a violin and one other taking part in a guitar. I stored attempting over and over once more. You’ll do one character taking part in a violin, and the opposite one’s sitting there or the opposite, or one character taking part in a guitar, and the opposite one’s sitting there. Nevertheless it’s too sophisticated to ask for it to do each, and I attempted each immediate I may consider.

Then, when it does it, generally the violin in seems to be too huge, or generally there are too many fingers on the palms.

Dan Holloway: Yeah, that is one thing I’ve come throughout loads. I’ve come throughout memes with this, with AI palms.

Howard Lovy: Proper, precisely. So, I truly specified as soon as, ‘5 fingers on every hand’, and I feel this factor has a character and is being sarcastic with me as a result of then it gave me an image with no fingers.

So anyway, there appears to be a substantial amount of trial and error, however I am studying methods to create completely different photographs, like generally it may be a cartoon picture, generally it could possibly be a 3D real looking picture.

My feeling is that it isn’t but changing an actual human artist. I feel from what I hear, the human artists, you create issues on Midjourney and then use Photoshop to deal with the remainder, like, have 5 fingers on every hand, for instance,

However it’s fairly superb to consider what it’s I am writing and then have it pictured, and then I can return and truly describe what I am seeing on the display, and that creates this suggestions loop, I assume.

Dan Holloway: That is what I have been basically utilizing it for, is for creating settings.

So, settings for libraries and cityscapes and issues like that, which it is loads higher at. It is fairly obscure and blurry, however you do not have these apparent, horrendous errors, like having seven fingers on a hand, and it’s actually efficient for that.

However I feel, and lots of people appear to be saying this, that precisely what you describe goes to be finally the way it’s used, that what you would possibly name the grunt work of making no matter it’s we will create could be completed by the AI, and then we as artists do the sharpening. So, it may possibly produce, within the case of a novel, it may produce 100 thousand phrases of textual content. It could possibly be the primary draft, and then we craft that into our voice and put our form and end on it.

Howard Lovy: Yeah. Properly, I used to be experimenting with textual content too, and we’ll get to that later within the present, however what I wished to say right here is that, what first modified my thoughts about this was my son goes to movie college and he got here house for a winter break, and he was telling me about how he is utilizing it within the inventive course of, and the way it’s not doing his work for him, but it surely’s doing lots of, such as you say, the grunt work.

It’s going to create an preliminary picture for him that he can then manipulate on movie to what he needs.

Dan Holloway: Yeah, precisely. So, it is saving us lots of time in order that we will truly spend our time doing what we’d name the excessive inventive worth stuff.

Howard Lovy: And that is to not say 10 years from now you push a button, and every thing’s completed for you. I do not know what the long run is, however I do know that it is right here and we both need to make it work for us or let it move us by, I suppose.

Dan Holloway: Properly, yeah, the way in which I’ve seen it put loads is, if we do not have a constructive or a hopeful outlook on it, then we’d as properly hand over, as a result of if we actually assume it will take our place, then what is the level of carrying on?

Writing a Youngsters’s E book Utilizing ChatGPT

Howard Lovy: That is AI relating to artwork, let’s go into writing itself. Within the information, you reported that somebody has written a youngsters’s e-book utilizing ChatGPT.

Dan Holloway: Yeah, so this was actually fairly fascinating, and it obtained individuals actually upset.

Ammaar Reshi, who’s a tech engineer, and I feel that is a part of what obtained individuals upset is that he’s a tech engineer and not a author, he is mixed ChatGPT with Midjourney to create a youngsters’s e-book. So, bringing the 2, the illustration and the textual content, collectively, clearly with his personal prompts.

It is brought about numerous anger, as I say, I feel partially as a result of he is type of seen as a little bit of a tech bro.

Howard Lovy: Are you aware any of the specifics as to how he did it? How a lot did he information the method alongside? From what I found, ChatGPT will solely do what you inform it to do to some extent, and then it goes off within the path you could not need it to go.

Dan Holloway: Yeah. So, what he did was, he began with a immediate, and then, I imply, I linked the Twitter feed that he is obtained in an article, he is obtained a extremely fairly fascinating video of the backwards and forwards between him and ChatGPT. So, he requested it for a immediate, it despatched again some textual content, and then there was virtually like a backward and forwards communication as he refined it, by no means taking it offline. So, he did not truly take the textual content and then put the ending touches on himself, however he obtained ChatGPT to change it till it was in some unspecified time in the future prepared. So, the phrase that he makes use of was, “it was like having a constant brainstorming partner who I could ping pong ideas off.”

Which truly, I like, and he could also be a tech engineer however that feels to me just like the inventive course of.

The Authorized Implications of Utilizing AI to Write

Howard Lovy: It looks like a youngsters’s story can be a very good begin. From what I found, its writing is uninspiring and juvenile in case you ask it to simply write one thing on a theme.

Now, in case you ask it to imitate your favorite writer, then it begins getting a bit extra inventive, however in a method that is clearly by-product. I feel it nonetheless solely produces what you inform it to provide.

Dan Holloway: Sure, completely, and this is among the controversial features that artists specifically, and clearly, it will be the case with writers too, have been upset about, is basically whenever you use it to create one thing, you then personal that factor, however the AI that has produced it has been educated on all these different items of labor.

So, in case you ask it to provide a narrative within the fashion of Roald Dahl for instance, or an image within the fashion of Vang Gogh, there are lots of these going round, or an image within the fashion of Picasso, or extra worryingly a residing artist, then the AI that produces that can have discovered how to try this by coaching itself on the work of that artist, however that artist then does not get any of the profit, I assume you’d say, of the ultimate business product.

Howard Lovy: I am unsure Van Gogh could be very serious about taking any credit score for mimicry.

Dan Holloway: No, however individuals are utilizing their favorite cowl artists, or their favorite illustrators, or their favorite manga artists and saying, produce me one thing within the fashion of, and these individuals are saying, properly, cling on a minute, it is utilizing my work to provide this picture, I am not getting royalties on it, and they’re then in a position to promote it.

So, that is the place it is barely blurry, and the businesses, that is the place they’re performing like tech corporations and they’re saying, properly, it is a bit bit too troublesome for us to ask everybody’s permission, these works are already on the market, they’re mainly scraping content material, in case you had been to ship us in one thing, then we might cease it utilizing your work, however that is as much as you to do it. So, there is a barely blurry dialog round how this AI is educated.

Howard Lovy: It’s extremely blurry as a result of I am unsure it is truly scraping content material, as a result of it isn’t taking it phrase for a phrase. For instance, within the story I am writing, it is about people who find themselves songwriters. I do not know methods to write songs, and that is purposely why I did it this fashion, as a result of I wished to see what the AI can do. So, I gave it a subject to put in writing music lyrics to, and it did it, and it is actually, quite simple, virtually juvenile.

However then I ask it to say, properly write a music within the fashion of Bruce Springsteen about A, B and C, and then it will give me a music that perhaps plausibly Bruce Springsteen may need written, or within the fashion of no matter your favorite songwriter is, and it will get a bit extra inventive, but it surely’s not something that they essentially would have written. I am unsure what sort of magic goes on behind the scenes the place it says, okay, this is a Bruce Springsteen music and this is how I am going imitate it.

Dan Holloway: I can completely perceive why individuals really feel upset, that they assume that they need to be getting a license.

When the UK authorities did its session on AI and copyright, one of many issues that ALLi’s responses was fairly clear on was that, in case your work is getting used for issues which are going to then have some type of business factor, then the individuals who generate that content material ought to be paid a license price for that work. So, that is in all probability what I would say on that.

The Potential Advantages of AI Generated Textual content for Indie Authors

Dan Holloway: However from a inventive aspect, what pursuits me is, if I had been to ask it to provide a piece on a selected topic, however in my fashion. So, if I had been to feed it my present novels, then I may see the way it thinks I might write it. So, that may give me a extremely fascinating foundation for then sharpening and tuning.

Howard Lovy: Proper, and I attempted that. I used ChatGPT and additionally Sudowrite. I fed it passages that I had already written, and I requested it to rephrase it, to see if it may possibly come up with a greater method of wording it, and it got here up with completely different phrases and completely different emphases and completely different phrases, however I am unable to say that it was all higher. It is simply completely different.

However in case you ask it to compose a scene from scratch, and you outline the parameters particularly, like, who’s within the scene, the place they’re, what temper the characters are in, it does not do too badly, and then you possibly can fill it in from there.

So, I assume the place we actually get into hassle is that if I say, properly write this opening scene within the fashion of an present writer.

Dan Holloway: Yeah, but when that present writer’s you, then would not that be fascinating?

Howard Lovy: Proper. Properly, I am not well-known sufficient for it to know what the fashion of Howard Lovy is, I’ve to inform it by feeding it the data.

Dan Holloway: Yeah, that is the issue with being a first-time novelist. Should you’ve already obtained a number of books on the market, and they’re all on open entry someplace, then I feel that may be a extremely fascinating method of, we speak about producing excessive quantity of content material, so that may be a extremely fascinating method of manufacturing excessive volumes of content material that had been authentically yours. You might get it to put in writing the 50-60,000 phrases, and you then would basically are available in on the editorial stage.

Howard Lovy: Proper. So, what about you? You are a poet and a author in several genres, how do you see your self or your course of being enhanced by ChatGPT?

Dan Holloway: I feel I am ready to have the ability to, I like the concept of utilizing it in that method, of claiming, these are my books, I need to write a e-book about, I do not know, caving. So, that is one thing that I’d conceivably write a e-book about, as a result of I like excessive sports activities, I’ve written quite a bit about different excessive sports activities.

So, I need to write a thriller about caving, which is a closed room thriller, which has a cheerful ever after ending, for instance, and please write it with my distinctive story growth and the emotional beats are the identical as they’d be in a Dan Holloway novel, and then see what it comes up with. I feel that may be actually fascinating.

Howard Lovy: Should you’re not an skilled caver, it may let you know what skilled cavers undergo.

Dan Holloway: It may. It may additionally inform me as a author what my ticks are that I may not concentrate on, for instance.

Howard Lovy: Oh, just like the phrases you repeat usually, issues like that?

Dan Holloway: Yeah, phrases I repeat that I’d do unintentionally. I used to be advised, I bear in mind the primary time I queried to an agent, the factor I used to be advised is I had a extremely repetitive sentence construction, I assume, topic, verb, object, topic, verb, object. So, ever since that my sentence construction has been in all places, but it surely was actually useful to listen to that.

So, I think about seeing it on the web page from phrases I have never produced, however I would requested it to provide in my fashion can be actually a great way of bringing that house.

Howard Lovy: There are different AI packages that may assist with that, I am excited about Fictionary. I signed up for an preliminary free trial on that, however I have never actually used it.

What ChatGPT does very well is analysis. I requested it, what did a music studio appear like in 1984? And it gave me, properly, again in 1984, there would’ve been a reel-to-reel tape machine, and that is what the management board would’ve appeared like, et cetera.

I assume, within the olden days, I might’ve gone to a library or requested a music producer what that may have entailed, however now I can ask ChatGPT. Or how do buskers set them as much as play music in New York Metropolis, and how do they get the allow to play in a avenue nook? Simply very particular issues that’ll find yourself particulars in my story.

Dan Holloway: The form of issues that we’re used to as writers, if we get it fallacious, then that is the factor the readers will level out to us.

Howard Lovy: Precisely. Proper. So, I see that as being very useful.

How is ChatGPT Affecting Corporations Like Google and Microsoft?

Dan Holloway: This after all is what’s obtained Google indignant.

Howard Lovy: Proper, and you talked about in one in all your columns that the Google has known as a ‘code pink’. I am unsure what which means, however some type of alert; our dominance is being is being challenged.

Dan Holloway: Yeah. Properly, from what I collect, it actually means, it is a potential existential risk as a result of clearly Google makes its cash from the adverts that go into its search. So, whenever you search, the highest gadgets are adverts, and there are adverts within the sidebars, however in case you’re not utilizing Google for search, then it isn’t going to have the ability to make that advert income.

So, in case you had been to ask it, what did studios in 1984 appear like, then it may also present you some adverts for, I do not know, synthesizers or audio interchange packing containers or no matter, as a result of it would determine that you simply had been actually serious about that, or it would present you an advert for Grammarly, which might be additionally going to be out of enterprise fairly quickly. Nevertheless it’s not going to have the ability to do this in case you’re simply going to ChatGPT to seek out this out as a substitute. So, it is having to do some severe excited about what this implies for its enterprise mannequin.

Howard Lovy: Yeah, and I am questioning, is Google arising with its personal model? Is it going to purchase it out? Will ChatGPT develop into the following Google, and will all of the adverts rush to that platform?

Dan Holloway: And that brings us to the story that is popping out on this week’s information, which is round Microsoft. As a result of no, Google is not, however Microsoft is, is the reply.

So, Microsoft is investing 10 billion into Open AI, which is some huge cash, and it is valuing it at 29 billion. So, that is a watch watering sum for one thing that is basically actually new expertise.

So, they’re basically going to be utilizing it, on the very least of their Workplace suite. So, as I’ve mentioned within the column for this week, it is basically an up to date model of the previous paperclip that all of us fondly bear in mind.

Howard Lovy: Proper, precisely, however solely this one can be a bit smarter than a paperclip.

Dan Holloway: Yeah. So, even writers who assume they are going to don’t have anything to do with it, basically the expertise that is going into ChatGPT goes to be a part of Microsoft Phrase in not very lengthy.

Howard Lovy: Attention-grabbing, and you talked about Grammarly is perhaps in hassle. Grammarly has a particular goal for many individuals, however do you assume ChatGPT would possibly change it?

Dan Holloway: I imply, grammar is a kind of issues that, you had been speaking about analysis, grammar is one thing we will analysis. What do individuals like to listen to in sure contexts? How ought to this sentence learn if I am sending it to my supervisor? How ought to this sentence learn if I am sending it to my date for the night time? How ought to this sentence learn if I am sending it to my agent?

There’s some extra nuanced variations of what Grammarly does with its contextual, that is how you must write an e mail, that is how you must write a proper letter.

Howard Lovy: Earlier than the present, I requested ChatGPT to put in writing an intro to this podcast. Do you need to hear what it got here up with?

Dan Holloway: Okay, let’s have a hear.

Howard Lovy: All I did was immediate it to put in writing an intro. I gave it our names and what the subject is, and it got here up with:

Welcome to the AI Creativity Podcast, the place host Dan Holloway and co-host Howard Lovy dive into the thrilling world of synthetic intelligence and its rising influence on the world of creativity. From AI generated artwork and music to AI powered writing and filmmaking, we’ll discover the newest developments on this quickly evolving discipline, and what it means for the way forward for human creativity.

Dan Holloway: That is fairly spectacular.

Howard Lovy: It is not too dangerous, it obtained the identify of the podcast fallacious, however I did not inform it the identify of the podcast, so it made it up.

Dan Holloway: It additionally actually fancies itself.

Howard Lovy: Sure, truly, if I do say so myself.

Earlier than we finish this, I ought to point out that we’ve an awesome weblog publish up on our website, AI for Authors: Sensible and Moral Pointers, and contributing there’s Orna Ross, Joanna Penn, Laurence O’Bryan of BooksGoSocial, Amit Gupta of Sudowrite, and J.D. Lasica of Authors AI, and all of them speak in regards to the moral and authorized implications of AI. It is a actually good learn, we’ll hyperlink to it within the present notes for this podcast,

Dan Holloway: Tremendous, yeah that is one thing we have not gone into, and clearly has been going fairly quickly once more, is the authorized place round it in numerous completely different international locations, and what copyright means on this planet of machine studying.

Howard Lovy: Properly, that may need to be a narrative for an additional day, we will speak about this for a very long time.

Properly, Dan thanks very a lot for this fascinating dialogue, and I do know that we’ll be speaking about AI into the long run.

Dan Holloway: Good. Thanks very a lot certainly.

Howard Lovy: Okay. Thanks. I am going to speak to you subsequent month.