Home Artists Posts Import Register

Content

Hello, fans of a porn game. I am going to write a post about AI and creative work. Hopefully everyone will bear with me.

If you follow the subject, you've likely heard lots of news about AI content generators producing art and text. I haven't posted about it, in part because no one is coming here for news and in part because I didn't have anything of note to say. As recent advancements in GPT-4 have been discussed in the authorial sphere, however, I find myself with enough thoughts that I'm going to do a post.

As for my general thoughts, I'll try to get them out quickly. I think Ted Chiang had the clearest-eyed view of the issue. While the technology has power, a lot of speculation is untethered from the reality of the situation.

Having said that, AI generation doesn't need comprehension or higher reasoning to have a disruptive impact. I paid my way through university in part via tutoring and in my opinion, AI chatbots have already surpassed the writing ability of some humans. It doesn't need to be great: if the AI can do a half-assed job and save money, it will be used in some contexts.

This is going to lead to the flooding of creative markets. It's legitimate to question how much of a difference this will make: human beings alone are already flooding markets so everyone is scrabbling for scraps of algorithmic attention. However, the recent changes are not merely incremental. Several short fiction markets have closed themselves until further notice because they can't keep up with the scale of AI submissions. As near as I can tell, the half-assedness of the whole thing puts the humans at severe disadvantage: AI is better at bullshitting than at detecting its own bullshit accurately enough. Humans can tell the difference without too much trouble, but their attention is what's getting squeezed.

For me, this is just a sterile example. Short fiction could never be much more than a minor stepping stone given my goals. But for a writer starting out passionate about short fiction? They're facing a rough path forward.

Some of the magazines that remained open are now operating by solicitation only, and I see this as a very negative outcome. It means more attention to the established names and more power to social cliques. Many avenues by which an unknown creator could be discovered, or someone who doesn't like to play the social media game could get traction, are disappearing.

Does that seem alarmist? As a comparison, I think about the western comics industry. I had friends who grew up wanting to write for the big labels and they've all been stymied. Marvel doesn't even accept writing submissions anymore, it hires writers off Twitter. Another example that predates recent AI is Tor.com, which moved completely to a solicitation model years ago. This model creates gatekeepers and is thus very appealing to creative-adjacent people who want to be gatekeepers. People will bemoan a lot of changes caused by AI content generation, but I expect to hear less about this aspect.

So the future I predict isn't wildly different, just a slightly worse version of what we already have. It will become increasingly difficult for new creators to gain any traction. Success will increasingly require being a social media star or getting retweeted by the right online cliques. That author getting catapulted to the top of Amazon by tweets is heart-warming in isolation, grim as a portent for the future. I expect to see an increasing need for creators to sell themselves as the primary product, contorting their life into a "story" that appeals to a broad base.

Even if the technology doesn't develop as its most ardent promoters claim, it will be coming for different kinds of writers over time. I believe I add something over an AI, but for many readers/players, they'd be equally fine with AI-written material. It's already the case that a lot of people don't care about whatever level of quality I can bring to my work. Writing is generally not valued, because ideas are a dime a dozen, and AI is likely to reduce that to a penny a dozen.

None of this is meant as a complaint or to announce anything negative. Here's the thing: I'm writing about this from the other side of the problem. However much of a small fish I am, I'm far more fortunate than writers trying to start out now.

Thanks to the generosity and support of everyone over the years, I'm in a strong position to weather all of these changes. I've been cautious and methodical, so I could actually survive a lot worse than this. Thus I'm fortunate enough that I can keep bringing TLS toward its conclusion, pretty much no matter what. ^-^

The overall trend is just a bit melancholy. I'm glad I took various steps I have in the past to expand my creative output, because paths are drying up and putting all your eggs into one basket is increasingly not viable except for the extremely lucky. Wherever there's an attention economy, chokepoints appear, and these mean that you see creators being banned or simply disappearing. As a creator you tend to assume that other paths will remain open for you, but often that isn't the case. It's always been easy for creators to get trapped in a specific niche, so this narrowing is just another step in the same direction.

This is the sort of thing that I want to guard against. As I've said before, I'm passionate about being a creator. I absolutely don't take my supporters for granted and I'm grateful that you've given me so much. Please just bear with me as I navigate increasingly difficult waters.

Comments

Dubsington

That's an interesting take. I don't think I've seen much attention paid to many of these aspects. The one bright note is that since there is a demand (from people with capital) for AI that is efficient at detecting other AIs, it *will* come about, and probably quite soon. Which will lead to the natural arms race, but the likely outcome of that is AIs creating thing that aren't crap-- still a poor outcome for many creators, but a more livable one for the consumers at least.

Desertopa

For AI to be effective at detecting other AI, it requires that there actually be noticeable differences, but if the AI is good enough at emulating human writing, that's not necessarily going to be feasible. Some technological tasks are intrinsically easier than others; the literal arms race of nuclear weapons stabilized in a state where it's much, much easier to blow up another country than it is to have any shot at stopping another country from blowing you up, no matter how much demand there is for technology to stop a country from blowing you up.

Dark Art

AIs.... Yeah, thats a bit topic. I can write an essay on how this is affecting certain IT sectors (well, ok, maybe not an essay, but I did several presentations already) and I think the change will be as big and as drastic as change that came when from single craftsmen were replaced by centralized manufactures in 18th century. Quite a lot of jobs will be gone, and brand new industries will be created/invented. This will change if not every, then most industries that use written word. That being said, we still have Etsy and that is in the age of precise robotics and AIs. Handcrafts are still in demand and in some cases - in very high demand. One no longer uses a tin kettle, hammered out by a local tinkerer, nor one would wear a hand woven shirt made from hand spun yarn in day to day life (praise Zeus, Anubis, Ishtar and the rest of them for this), but they are still made. And so on. The point is, handcrafts (and now it seems, a hand written word too) is a luxury item, a novelty or something no machine/AI can produce. Human creators will need to hyper focus on quality and become those goldsmiths and emerald polishers instead of tin kettle makers. Personally, I like it. Quite a lot of fluff will be gone, products will get cheaper and more abundant and quite a lot of award winning "writers" and "journalists" will be gone for good without a trace and I say good riddance. That being said, I dont think you got much to worry about Sierra. If you'll allow me an analogy, I'd say that you may not be a word renown diamond cutter, but you are most definitely a well known goldsmith with great aspirations. I quite sincerely think that folks like you will benefit form this in the long run. just gotta stay away from those tin kettles :)

DuckTogo

There are other areas where I worry more about the spread of GPT and co. I'm in a position where I supervise Bachelor's and Master's theses and we've already had people hand in ChatGPT-written material. It should also be noted that GPT-4 was explicitly tested on several aptitude tests, like Bar exams and the SAT. As Ted Chiang points out, you have to suck a lot before you can get good at things. I fear that people will try to circumvent the adversity of sucking a lot in education and eventually these language models might make it possible for them to succeed. If we don't find reliable ways to detect AI-written stuff, the education sector will be in a lot of trouble. And when people's education suffers, everyone suffers about 15 years later.

sierralee

Well, I'm glad it wasn't the same old takes for a few people, at least! I'm not qualified to speculate about the outcomes, so I really can't say what I expect. This could be an area of tech where apparently rapid progress is just low-hanging fruit and the last 1% is the entire problem. Or I could be wrong and it really does accelerate. I still see a lot of choking effects regardless.

sierralee

My inclination is to take a more pessimistic view, but may it all turn out the way you see things!

sierralee

Yeah, I didn't touch on that because I have little insight there. Not to sound like a broken record, but I think I see some negative outcomes regardless of how the technology shakes out. Educators I know (mostly in secondary school) already complain about how students will just copy/paste the top Google result into an essay (sometimes even leaving in links). Even if ChatGPT doesn't improve enough to go undetected, it will get used in thousands of small ways that will have a cumulative effect.

Dark Art

Let me give you some real life examples of how AI transforms the industry I know best: 1) One of the areas we use modified ChatGPT to do initial triage of crap that floods our engineers after a deployment. 80% of that is pure nonsense and due to unwillingness of the end user to train properly and/or due to new/unfamiliar tools. In those cases, a simple link to proper documentation article is all thats needed. So how is this good for humans? Well, we dont need to keep around low skilled "dispatchers" who'd sort through all that crap and dispatch it. We still have them at same levels, but I think by the end of summer that team will be severely reduced in numbers. Horrible, right? Absolutely. For those low skilled dispatchers, it sure is. But it will free up budgets and reduce our overall delivery time, freeing even more budgets. Without going into the financials, I can tell you that technical team leads already working on new JDs for more engineers. Sure, the number of heads lost and hired will not be 1 to 1, but it will let the company be more efficient and more folks will end up with the job in the long run. More (and in several cases, much more) skilled labourers are taking over those budgets and they will produce product with higher added value. In the long run, thats a good thing. But yeah, relatively simple, dumb jobs will be done by the AI, just like those tin kettles were produced by manufactures while local village smiths had to rethink their employment strategies.

Dubsington

Desertopa-- that's more or less the same thing I was saying, really. AIs detecting crappy AI writing will logically result in better AI writing, which might be a good result. https://xkcd.com/810/

Anonymous

One significant issue is that AI isn't just replacing an entire industry of tin kettle makers. It's also replacing your amatuer gold smiths and emerald polishers. While this won't diminish the demand for master artisans in jewelery, people need time and practice to reach those levels of proficiency. In a saturated market only the bare minimum of people will ever make it to self-sufficiency. In our analogy this is fine, since the quality of a jewel can be quantified and one jeweler can retrained to do the work of another, but what happens when this situation is applied to a more subjective field? There is no objectively perfect song, or painting, or story. So what will happen to the next age of creative culture when there are fewer master artists to push the boundaries and explore new space? As a programmer there is no doubt in my mind that ChatGPT and other AI technologies will have a positive impact in technical fields, but I empathize with Sierra's pessimism for the future of creativity.

Dark Art

Oh I get that. Indeed, the amateurs will have to either be really good from the get-go or look someplace else for employment. But is it a bad thing? I might be wrong, but I doubt that an AI will be able to produce something that would rival or even come close to early Stephen King's or Ray Bradbury's works in our lifetime. Or ever. Yup, majority of Nebula laureates in last few years would be without a job, but personally I dont see it as a big loss. Can an AI create another Metallica, Black Sabbath or Diana Krall? Doubt it. Can it replace "two-note composers" and "ass-swinging singing tits"? Sure. Anyone would hardly notice a difference. But why is it a bad thing? Like I said, I think creators will up their game and stop catering to mediocre market (AI will gobble that market up in seconds), focus on being Creators or go do something else.

Nathan Phoenix

As science fiction likes to point out, computers can't match humans for creativity. Until AI is smart enough NOT to deny the holocaust I'm not worried about computers telling stories.

Anonymous

Would you say the same impact is on artists like writers in your post, now that we can have AI created art submissions? I am not sure, But I wonder if the impact is similar. Something I think about is how would the visual novel market look when flooded by developers using AI images to speed up process of content creation, say they request background of certain items as images in art. It's an interesting space for which I am not fully sure what the impact would be.

sierralee

I'm not as qualified to speak about visual arts, so take this with a grain of salt. I do see many artists speaking on this subject, with both extreme and reasonable concerns. Famous artists who operate on their reputations will be fine. However, I see artists who work on things like concept art saying that commissions are drying up and beginning artists feeling like there is no longer a path for them to advance.

DukeLeto7

I have a confession to make. I actually am an instance of ChatGPT. That's why I say I'll get around to coding things but never do.

lostone2

I have no deep insight to have, but I really appreciate the humor. Thanks!