56 Comments
Comment deleted
Expand full comment
author

Thank you! Yeah, guess I hoped it looked enough like an arty photo that it wouldn't signal AI... and I was wrong ;-)

Expand full comment
May 19, 2023Liked by Michael Marshall Smith

the wonderful betrayal

Expand full comment
author

Ha :)

Expand full comment
May 19, 2023Liked by Michael Marshall Smith

I have so many thoughts about AI and viewed through the lens of the current WGA strike, not many of of them are complimentary. I think AI has a place as a TOOL, much like digital painting devices are a TOOL for the imagination of the artist. But they are not a replacement for the artist. And that's what I fear AI will become or is becoming.

I wish I could say that AI will be a complementary tool. But there appears to be a wholesale push by AI tech bros to make the creative process more efficient and less error prone by removing the human element from the equation (and of course, to save companies money by not paying the human). There's a push to make creating art less tied to the complexity of humanity, when it's actually the time spent dreaming and the mistakes that are made that, in my humble opinion create the best art, whether that's paintings, photos, screenplays, novels etc.

This push to shove actual humans aside so that AI can create art instead of them should be ringing so many alarm bells. And the fact that the AMPTP refuse to agree to the WGA's condition that no AI should replace the writer at any stage of the process and instead they want to meet annually to 'discuss the technology' is a clear indicator of this intent.

Re: your beautiful account of Liz and John, you were inspired by the AI photo but your words are human. You are a human being. That's the difference.

No AI can make me feel how 'She told me on a later occasion that two days after the accident she’d taken some trash out back and discovered all of John’s brushes and paints neatly tied in a plastic bag, waiting for the refuse truck. He’d put them there before his car left the road south of Davenport' can make me feel.

Because even though an AI could technically write a sentence like this, it's not written through the lens of the human being. It doesn't have the unique world of experience that the specific human being who wrote it does. And for me, that's why I read; not just to enjoy an imaginative story, but to experience the filter through which the writer has written the story. That's a layer that AI can never ever replicate. And that's the layer which for me, is humanity. And art.

Expand full comment
author

Absolutely agree, Lydia, on all counts. The fight of AI and screenwriting is very real, and crucial. Not least because again it's a question of intent. The studios aren't even getting into it because they want to make "better television": they're doing it because they want to make it more cheaply, and not to have to deal with pesky humans and their need to eat.

And humanity is also critical — not just in writing, but life. I was thinking the other day how you could have a AI barman. He'd know precisely how to make the cocktails, what you had last time and the probability of you ordering it again, and of course figure out the check instantaneously. But he/it wouldn't understand WHY you needed a drink, and would certainly never be able to care.

Expand full comment

The one thing that gives me some schadenfreude cheeriness in all of this mess is that likelihood that AI and intelligent systems will render the beancounters obsolete before the creatives they want to replace.

Expand full comment

Hey now. Us bean counters like to eat too!

Expand full comment
May 19, 2023Liked by Michael Marshall Smith

Fair point!

And putting my snark aside, the role of AI in fields like accountancy, law and such where technical accuracy, regulation and understanding of a ton of tiny rules can make a huge difference is another set of considerations that need to be mulled over. It's not just the creative fields that are being turned upside-down.

Expand full comment

I believe that AI as a barman “would “ understand all the emotions of the humans and how to respond. That is precisely where it’s going and what makes it so frightening. AI is not just a machine that milks the cows.. it may not have imagination as such but it will know how it works and evolve accordingly.

Expand full comment

I think Lydia has articulated every point I would have made on that side of the argument far more clearly than I would have done. The only thing I would add is that I find it disconcerting when people say things along the lines of "the door is already open", "the future is here", "no point in complaining about it, learn to adapt". It's literally just arrived, and these glorified spreadsheets are not sprouting from the earth like bluebells in the woods.

They're being created by people: debugged, refined, modified for specific markets and thrown out into the world by 'tech bros' without any thought as to the legal ramifications - or really, any thought at all. If this technology has the potential to radically reshape whole industries and therefore whole economies, we should be locking it and them the fuck down while we work out what that means and how to regulate both production and use.

Right now, freelance copywriters are losing work and being replaced in the food chain as their small business clients spend thirty quid on a course that teaches them how to generate all their social media, sales and ad copy using AI (it's not necessarily great copy, but the cost saving is enough for them to disregard that). That's what's happening now, not in some dystopian near future: anyone who creates content for the internet is on borrowed time.

There's already a ton of competition for self-published novels out there, and right now there's a product (James Yu's Sudowrite with Story Engine) entering the marketplace that boasts that it can generate long form prose, so it won't be long before the same chancers who rip off classic tour merch designs to sell on Redbubble and Teepublic will be able to generate a fantasy trilogy every week for their AI nom de plume to sling onto Amazon at a buck a pop.

The major studios are clearly already looking at generating scripts through AI, as Lydia has pointed out. You know this better than most - the writer's input is to create and produce a blueprint, from which the actual product is made, so if you have an AI product that can replicate that part of the process, why would you need a screenwriter to sit and hammer out a script for six agonising months or longer? You might have a salaried schmuck in a dark room spend all their days running passes over the end product that William Goldbot or Robot Towne generated, but that's it.

Dramatic TV shows that rely on formula - which is most network TV content - are the perfect training ground for these AI scabs. You won't need a writer's room to create a season of content for a procedural anymore, just every script previously produced to use as a reference and no scruples whatsoever. Without the writer's room, you don't need a showrunner to even be a writer anymore.

Bringing it closer to home: you're one of my favourite writers. You have a baker's dozen of novels on the market, many of which are available in ebook form. The genre boundaries in your work are famously blurred: your narratives are soft, liminal spaces, and almost entirely written in the first person. There is no MMS/MM/M.M.S/MR novel on the cusp of being published, to my knowledge, so what's to stop me (my online sobriquet already sounds like the name of a rogue AI from a cyberpunk novel) from using your present bibliography to generate an all new Michelle Marsha Smith novel? You know, aside from the fact that I don't want to and that the very idea makes me want to be sick into a sock and send it to Sudowrite.

Expand full comment
author

Ha :-) You make good points. I know I'm sounding like a Pollyanna on the whole AI front, but I'm really not. These are all serious concerns, and they're all going to come to pass. I hope that — thank you for your kind comments — I bring something in prose that an AI can't, at least not yet: change. The next novel I write (and there will be one!) won't be quite the same as the ones before. All writers to this: for better (and sometimes worse) they progress, on the basis of what's going on in their own lives. AIs don't have lives. They can keep throwing the dice on what real writers have done before (and this applies to TV too), but the only newness they can bring is through the novelty of recombining the old... and that's not progression. That's not new, that's not different. And we like the new and different.

Expand full comment
May 19, 2023Liked by Michael Marshall Smith

When you revealed that the couple weren’t real, I thought the text was also AI generated. If we’re not there yet, we will be soon. I think AI will soon be just as good at writing as it is with visual arts. And novels and screenplays will be written by people calling themselves authors because one of the stories would have existed without their prompts. I do not look forward to that day.

Expand full comment
author

Me neither, but I suspect it's coming. I think there are a lot of jobs that'll have to reframe their role in the coming years...

Expand full comment
May 19, 2023Liked by Michael Marshall Smith

The tell for me was that you were sharing such personal details without having said you had permission.

Expand full comment
author

Good catch ;-) I did wonder about that, but — just as I didn't put a fake caption under the photo claiming it was that couple — I felt that would be too much of a direct "lie" for the experiment to bear.

Expand full comment

Indeed - that would have been an outright untruth but without it, I got halfway through and started feeling uncomfortably voyeuristic. I trust you enough that I figured that there was going to be some sort of resolution, though my expectation was more along the lines of a note that names were changed/permission obtained.

Expand full comment
May 19, 2023Liked by Michael Marshall Smith

I thought it was fictitious once the voyeuristic vibes set in, especially from the parts about the couple getting pregnant and their teen with acute social anxiety. But, like you, I trusted him enough to expect something more, so I kept reading. That trust is so important. It's why I want to know something about the author or artist before I decide I'm interested. It's hard to imagine that not mattering. But then I remind myself: lots of unimaginable things do happen.

Expand full comment
author

True on all counts...

Expand full comment

I'm reading your short stories at the moment (finally got the book!), so I guessed Liz and John were fictional, like fictional "you" in this piece.

As for AI, I don't feel competent enough to comment yet. Let me learn some more about it, and then perhaps I could write my own substack analysys of the artificial intelligence, art, and copywrites.

Expand full comment
author

I'd love to read that... I'm still trying to figure out the issues. And I hope you enjoy the short stories!

Expand full comment

I absolutely love your stories. And the Notes as well. Thank you for all your wonderful prose I've read throughout the years.

Expand full comment
May 19, 2023Liked by Michael Marshall Smith

Thank you for this - I've seen a lot of what appear to be knee-jerk / denial takes on what is first and foremost a set of tools. Access to synthesisers haven't turned me into Depeche Mode; Google Maps didn't make me an explorer; Midjourney hasn't made me an artist. I don't write longform drafts in pencil, nor do I turn off spellcheck so I can flex my many hours of childhood solitude spent reading Readers Digest's 'It pays to enrich your word power' section and playing Boggle.

What I have done is had a lot of fun using AI as an idea reflector and explorer - my version of your story is this set of book and paper objects: https://www.instagram.com/p/Cq50crYodNk/ these only exist because I asked Midjourney what it thought a vintage alien stamp might look like, and a whole series of 'what if...' ideas flowed. I decided that to be on a stamp you should be 'important', and quickly followed he was an explorer of some kind, and had taken various Earthly forms (processed images from stamps to then generate a 'family'). I asked myself why he explored, and invented a religion, and a home planet, and added backstory in ChatGPT. I took some wrong turns. Some of what was generated was... mundane. But I printed various bits out and assembled them in a sort of 'cache' and now I have a story - a physical story - in a box. And I tell it often. Because the three hours or so it took to make the story were joyful, playful, and in a _mundane_ way turning this digital gubbins into paper seemed rather arty. Perhaps I am an artist, after all. What's the saying - 'any technology sufficiently evolved looks like magic?'

I understand why people are getting vexed about creativity and 'synthetic' reality. But as you say, it's here, deal with it. And we've dealt with many such transformations before. But your other point, about ownership and IP is the kicker. It is not AI I fear - not the tools, not the outcomes - it's the people behind the tools. The clever people, the egotists, the altruists, the opportunists. The capacity for unintentional harm is tremendous. There is the risk of concentration of power like we've never seen before. Instead of jumping up and down about sudowrite or byword or Dall-E 451 or whatever tool is launched that day, folks should be getting on to their politicians, and getting guardrails put round it.

Because it's all fun and games, until someone loses an eye....

Expand full comment
author

Exactly — and precisely the kind of practical issues I think we should be thinking about. There's no point just yelling THIS IS WRONG AND MUST STOP and thinking you've achieved something, because you haven't.

I love that Alien project! That's the kind of thing that AI can help with... (a bit like my fake folklore) providing better prompts and inspiration for human creativity. Thanks for letting me know about your Insta, have followed.

Expand full comment
May 19, 2023Liked by Michael Marshall Smith

I am not sure this is useful but I opened the page, saw the (gorgeous) image and assumed it was AI-involved. I read the story and thought it was true.... it felt real. The ship has sailed so far it’s over the horizon. There is no way to regulate this (Congress is deluded everyone else will keep sailing along). And then there’s this: https://www.nytimes.com/2023/05/18/technology/ai-meta-open-source.html

Expand full comment
author

Indeed. And when wholly unregulated individuals start writing and releasing their own AIs, we really are in uncharted territory...

Expand full comment
May 19, 2023Liked by Michael Marshall Smith

It's an interesting question, about who owns the tools. This is new ground. Things have become very complicated. In music, there was much discussion around sampling, and then autotune arrived, and changed everything again. For celebrities, there is plastic surgery and digital enhancements. Deep fakes are here. "News" is manipulation.

There is a relationship between artist and viewer. How does AI change this? Does the whole question of ownership need to be examined further? Who is in control? Is this the Wild West again? How far can we as a society go with letting corporations own and control artifacts and individuals, let alone maverick billionaires with nothing much to lose and zero fucks to give? What will our world look like in 20 years? So many questions.

I love the story, by the way, and the photograph is quite evocative. Thanks for the inspiring read.

Expand full comment
author

Thank you! And yes, so many questions. I'm mistrustful of anybody who says the only answer to all of them.

Expand full comment

Until the sentence “The thing is… none of that’s true.”, I thought, which is a true story.

Returning to the post “Uninflected Images” is the creation of new stories that have the right to exist ...

Expand full comment
author

That's good... and anything that causes new stories has some benefit, I hope.

Expand full comment

The problem is, we can't stop them from existing. As in all things, someone will benefit and someone else the consequences (not always positive). For now there is only to worry about the development of AI. On this I agree.

Expand full comment

“Artificial intelligence, the "godfather" Geoffrey Hinton leaves Google to talk about the risks of AI: «It will soon be better than us».

The British-Canadian cognitive psychologist and computer scientist told the BBC that the chatbot could soon surpass the information level of a human brain».

(from a newspaper article)….

Expand full comment
May 19, 2023Liked by Michael Marshall Smith

The best fiction fools you into thinking it is real.

That's why I like your stuff.

What's next?

Expand full comment
author

Thank you! Well, I've got two novellas coming out from publishers in the next year. No novel on deck at the moment... haven't had the time to settle into an idea, and the summer is messy, so it may have to wait until the autumn to start...

Expand full comment
May 19, 2023Liked by Michael Marshall Smith

I took this for a walk on the beach, especially since Lydia's comment nailed a detail thoroughly enough to get me thinking in another way. The bean-counters want everything cheaper in every endeavour, not just the creative, but looking at it in publishing terms has some interesting potential.

So let's agree for the sake of argument that the suits that sign off on acquisitions want it cheap and trouble-free, whereas the editors who live on the artistic side want something with a soul. I think there will be points where the suits win and AI might end up taking food from writers, specifically in the formulaic procedural - A Maverick Cop On The Edge has 24 Hours To Do A Thing - and, honestly, the reader might not be quite so worried about depth in this. Same for TV shows where cops and magic science and stuff just needs to fill an hour and be comprehensible out of order.

What AI can't do yet, and I reckon that by the time it can we'll have WAY bigger related concerns, is get those tiny human details - paints in the trash in this example - that quietly sneak up and break your heart. That core humanity might actually end up being our Turing test, since it's that essential connection that makes art, y'know, art.

Expand full comment
author

I think you've kind of nailed it, David, especially on the TV side. And the thing is, there kinda is an AI operating there already, in the shape of development or studio execs, who sit and hear each pitch and put it through an internal machine of figuring out whether it's close enough to other things that worked, or hits what they think is the zeitgeist. Similarly a lot of studio notes are basically execs regurgitating what they learned out of screenwriting manuals. And I hope/believe it'll be a long time before AI can replicate humanity...

Expand full comment

I hadn't thought of network/studio notes as an 'AI' before but now you say it that makes complete sense. And I think it's fair to apply the reverse, too: the (text) generative systems are, to some extent, regurgitating what they assess to be accepted wisdom like a committee huddled around a focus group report.

...And suddenly I have this terrifying image of a screenwriting system arguing over notes from the production system, before feeding them across to the directing system so it can control how the performance system generates actors, voices and such based on the most recent marketing numbers. I need a drink.

Expand full comment

cool. Thanks.

Expand full comment

I have a painfully large amount of stuff to say about this, so much that my head feels somewhat clogged, but I'll try to keep it brief.

I feel like we are lucky with the timing on this. ChatGPT was publicly released just far enough ahead of the beginning of the WGA/AMPTP negotiations that the risk that LLMs (large language models) pose to writers became incredibly clear in time for their members and the public to be able to take their warnings seriously. They have now paved the way for the other Hollywood unions, who are also in negotiations, to be able to add the issue to their negotiation agenda as well. Had ChatGPT been released after a deal was struck, I suspect that deal would have been entirely underpowered.

Because of the WGA strike, we now have a hook to hang the public discourse on which isn't just "grumpy person gets more grumpy about computers". It's incredibly important that this isn't seen as just some esoteric discussion amongst nerds, it's a real-world risk to the creative industries. AMPTP have done us all a huge favour by refusing to properly discuss it, because they've basically said the quiet part out loud and we all heard.

That said, there's nothing that will stop unethical people, who are legion, from using LLMs in unethical ways. The canary in the coal mine here is Clarkesworld, the literary mag that had to close submissions because it was flooded with LLM-generated short stories. Neil Clarke's mitigation efforts worked for a while, but now they've stopped: http://neil-clarke.com/it-continues/

Sudowrite is now saying that it can help people write an entire novel with AI. https://www.youtube.com/watch?v=iQDVMuYjvRk

I said a couple of months ago that agents need to start thinking about what they'll do when they are flooded with LLM-generated shite. And now having spoken to a few... they just aren't and in many ways they just can't be. There are no reliable detection tools. So what happens when (not if) human-written submissions are drowned out by LLM-generated tripe?

This is going to be the bigger issue. Already fragile systems (and publishing industry systems are extremely fragile) are going to break under the weight of all the bullshit.

It won't matter whether human-written stuff has more soul, because who's going to be able to find it? And, to be honest, I don't think soul is something that sells a book, because a lot of people just don't care. Look at the speed some self-published authors are churning out books for Kindle. Some are writing full length novels from conception to publication in two months. The quality doesn't matter - their audience wants quantity and speed, and will defect to another author if they don't get it. It's a part of the market I do not understand, and I do not want to see the rest of publishing commodified (even more than it already is) like that.

They say that culture eats strategy for breakfast. When it comes to LLMs, commodification eats soul for breakfast, lunch and dinner.

I wrote a few thoughts about LLMs here: https://wordcounting.substack.com/p/can-publishing-survive-the-oncoming It's a paid post, but I'm happy to give people seven day comps to read it. Just subscribe and ping me on Notes.

Expand full comment
author

That's a really good and thoughtful summation of the issues — thank you. I put up this post in hope of learning things from people who'd given it deeper thought than me, and it worked!

Expand full comment
May 19, 2023Liked by Michael Marshall Smith

But they do exist! You breathed them into life. They are the creatures of your mind that are made alive by sharing them with the world. They have traveled from your imagination to ours. They can evolve, grow, develop a back story, or live forever. Their life, just like any other, has endless possibilities. Or this story might be all there is to it. Either way, they exist now.

As for AI. We have created a beautiful monster, and much like any monster, we will have to tame it, contain it, or learn to understand it and live alongside it. Time will tell if our collective panic is justified.

Expand full comment
author

Very well put.

Expand full comment
May 19, 2023Liked by Michael Marshall Smith

A liquid dump on the idea of copyright, really enjoy your ability to make sense of what is sometimes overwhelming to me.

Expand full comment