We are literally at ground zero of the potentially transformative explosion. There’s so much discussion of AI and so much of it is bad, so much is emotional. Literally every aspect of AI is just… insane! I thought maybe to do a really good panel, we should start bouncing ideas and positions around in advance, take some time to think about things. It’s a huge amount of ground. People feel strongly about this topic, and rightly so. This is not business as usual.
IS AI A BUBBLE, AND IF IT IS, WHAT HAPPENS WHEN IT BURSTS?
* Obviously – it’s bad for everyone.
* This is likely to hit the mainstream publishing / bookselling industry, and likely contraction.
* This is likely to hit the mainstream publishing / bookselling industry, and likely contraction.
* Are there discernible ways it will affect writers and artists?
* Massive amounts of scraped data, including everything from work under copyright to private medical and psychological records are going to be assets going into bankruptcy/receivership – what happens to it after that?
* What happens after? Is it cheap discount chips and data centers for everyone?
* Will there be a next generation of AI? Cheaper? Rationed? Metered?
* Will there be a next generation of AI? Cheaper? Rationed? Metered?
WHAT ABOUT ALL THE SCRAPING? THE SCRAPING DONE. THE SCRAPING PROMISED? SCRAPING OF G-MAIL, OF THERAPY RECORDS, BUSINESS RECORDS, SLIMY CONTRACTS? THE LOSS OF PRIVACY, OF INTEGRITY?
* Are we fucked, or what?
* Is there redress? Is there a future?
* Or does it just go on?
* Or does it just go on?
WHAT ABOUT COPYRIGHT AND AI OUTPUT?
* Currently no copyright – but will that be revisited?
* How much human involvement will ground copyright?
* What about recognizable underlying copyright – is this AI a laundering operation, or is there liability?
* Is copyright meaningful – some slopmaster generates two hundred books a year, the revenue from accumulated volume is such that copyright just stops counting for anything. Are we seeing the end?
* Currently no copyright – but will that be revisited?
* How much human involvement will ground copyright?
* What about recognizable underlying copyright – is this AI a laundering operation, or is there liability?
* Is copyright meaningful – some slopmaster generates two hundred books a year, the revenue from accumulated volume is such that copyright just stops counting for anything. Are we seeing the end?
HOW ARE THE EVOLVING ECONOMICS OF AI GOING TO WORK OR APPLY TO WRITERS
* The original model has been ‘free use’ to get people to sign up / get addicted;
* Obviously, this has lead to haemmoraging money, which is continuing.
* The currently evolving model is ‘metered’ AI – AI charged in terms of ‘tokens’ or ‘hourly use’ and programmed rates.
* This is not currently effective, the costs per hour/usage far outstrip the hourly charges/tokens, which suggests the model will reform.
* Possibly costs per token hour/use will have to rise exponentially, but there’s a limit to what the consumer market will bear;
* Possibly costs per token hour/use will fall dramatically as new generations of AI become cheaper until its viable;
* Possibly we’ll see introduction of ‘stupid 2nd rate’ AI poor performing but cheap so as to be viable;
* Possibly a shift towards institutional clients/institutions that can bear the costs;
* There may be a realignment and gulf between people/parties that can afford AI and are motivated to use it expansively, even predatorily, on mass scales – forced to do so by the economics of the expenses; and those who can’t afford it and are shut out.
* Where do writers fit in each scenario? Where does publishing fit in each scenario?
* Would we see a scenario where writers vanish, and publishing and books become the domains of Editors as front line workers processing large volumes of AI for publishing houses which use their economics of scale and market positioning to dominate?
DOES WRITING/PUBLISHING AS WE KNOW IT HAVE A FUTURE?
Points to consider
* AI Slop is currently overrunning online nonfiction, including clickbait features, sites like medium, quora, and even youtube – not likely to decline;
* The number of books published has doubled or tripled in the last few years, apparently mostly due to AI, this is likely to continually increase, the deluge may be only beginning.
* We cannot count on quality of work distinguishing us from AI – it is inevitably going to get better.
* We cannot count on quality of work distinguishing us from AI – it is inevitably going to get better.
* The production capacity of AI is literally hundreds of times any human capacity to write – take a year to write a book, AI can pump out a hundred or a thousand shitty books in that years.
* Copying any style, riding any and every trend, kind of like the way Asylum pictures rides big budget Hollywood with low budget mockbusters, so basically, able to literally instantly prey on any active niche.
* So far, exponentially cheaper – which means it can be sold exponentially more cheaply, and not only flood the market, but collapse the prices.
* The historical precedent is that every time a crude technology appears which outproduces (poorly, but cheaply) craft or art, then craft or art inevitably collapses as a viable economic activity. Is this us? If no, why are we special?
* So far, exponentially cheaper – which means it can be sold exponentially more cheaply, and not only flood the market, but collapse the prices.
* The historical precedent is that every time a crude technology appears which outproduces (poorly, but cheaply) craft or art, then craft or art inevitably collapses as a viable economic activity. Is this us? If no, why are we special?
* One notable counter-example is painting, which in the face of photography, continually evolved, finding new styles – abstractionism, graphic novels etc., and building new cultures, where it remained viable -basically painting retreated or moved to new grounds and forms that could not be depicted photographically. Does this hold hope? Do we lose the territory we occupy, but can we find new niches, new forms and forums of writing, new markets and cultures that AI will not sustain?
* How much of what we’re doing now can we keep doing? How much of the current landscape will remain in place? If any?
* And if there is a future, how do we find our way to it? What’s the next step?
* How much of what we’re doing now can we keep doing? How much of the current landscape will remain in place? If any?
* And if there is a future, how do we find our way to it? What’s the next step?
THE HYBRID MODEL OF AI USE?
* Currently some writers are experimenting with a hybrid model.
* Use AI for the business/background aspects of writing, but not the actual writing itself.
* Background uses include research (HAHAHA) (acknowledging it’s shitty of me to laugh at AI research, I don’t care), marketing and market analysis, writing taglines, cutlines, summaries, synopsis, blurbs, doing proofing and editing, did I miss anything…..
* Given that AI is built on massive theft of writers work, is there an ethical paradox here? And if so, how do we manage it?
* Is it possible NOT to use AI given the way it’s being shoved into everything? Do we have that option?
* Can we afford not to use AI, and remain competitive in the field, or have any kind of chance?
* If we have to use it, what are the ethical lines? What’s an ethical use, as opposed to an unethical use?
* Does the hybrid model have a future, or is it transient? Will our efforts to use AI within boundaries mean anything, or will the tidal wave of slop make us irrelevant?
* Background uses include research (HAHAHA) (acknowledging it’s shitty of me to laugh at AI research, I don’t care), marketing and market analysis, writing taglines, cutlines, summaries, synopsis, blurbs, doing proofing and editing, did I miss anything…..
* Given that AI is built on massive theft of writers work, is there an ethical paradox here? And if so, how do we manage it?
* Is it possible NOT to use AI given the way it’s being shoved into everything? Do we have that option?
* Can we afford not to use AI, and remain competitive in the field, or have any kind of chance?
* If we have to use it, what are the ethical lines? What’s an ethical use, as opposed to an unethical use?
* Does the hybrid model have a future, or is it transient? Will our efforts to use AI within boundaries mean anything, or will the tidal wave of slop make us irrelevant?
COGNITIVE ISSUES
* Reading and attention spans have been in decline, will AI bring about a collapse?
* Will we lose readership, simply because no one reads books or stories?
* Is this the end, or are there new forms and formats – like flash fiction or other forms of interaction? If so, what’s out there for us? Any guesses?
* Is the novel going to be obsolete as a form (actually, this question applies to different categories – including business and market)
AI AND FRAUD
* Does AI open the door for new waves of fraud? More and better writing scams? More effective targeting? Fake services? Piracy?
* What does the future hold?
&&&
OTHER ISSUES
* General considerations of – Is AI a bubble? It does look that way. But what does it mean. If the bubble pops, is there anything left?
* Previous bubbles have left all kinds of infrastructure to be repurposed, from Fiber Optic networks, to railways. But in AI… basically the chips have a limited lifespan and burn out fast, what fluidity is there to repurpose, what’s the usage or repurpose of a data center? What are we left with after the bubble bursts, if it bursts?
* What does a post AI bubble world look like? Will we still have slop? Will we still have our data, including our writing, stolen?
* AI this is promises to be a kind of pervasively transformative technology which bills itself as affecting literally every aspect of life – and that it is completely unregulated, almost completely unstudied. We are Generation Lab Rat. It’s taken us years, even decades to assess social media and understand its toxicity, we’re not even bothering to look at anything. Every social and sociological and economic and cultural effect – it’s just going to be loaded on us, and we’ll just count the bodies later. I find that insanely reckless. Imagine building airplanes or nuclear plants that way?
* On this front – is AI even compatible with democracy, or is the potential for AI to be used to manipulate so pervasive and profound that its literally game over.
* What does a post AI bubble world look like? Will we still have slop? Will we still have our data, including our writing, stolen?
* AI this is promises to be a kind of pervasively transformative technology which bills itself as affecting literally every aspect of life – and that it is completely unregulated, almost completely unstudied. We are Generation Lab Rat. It’s taken us years, even decades to assess social media and understand its toxicity, we’re not even bothering to look at anything. Every social and sociological and economic and cultural effect – it’s just going to be loaded on us, and we’ll just count the bodies later. I find that insanely reckless. Imagine building airplanes or nuclear plants that way?
* On this front – is AI even compatible with democracy, or is the potential for AI to be used to manipulate so pervasive and profound that its literally game over.
* Instead of a publicly owned or controlled resource, with at least some accountability, the model we have for development of AI is for it to be privately owned by what appears to be a collection of narcissistic sociopaths. Even if the field wasn’t dominated by raging lunatics, the nature of any corporation is legal sociopathy – a corporation cannot legally have any other goal or objective than increasing shareholder value. Personally, I find that terrifying.
* What about privacy – people suggest using AI for therapy. Well all those therapy sessions will remain permanently as ‘training data’ just because that’s how AI works and are accessible to anyone with privileges or a knack for prompts, or who maybe just buys it. Every activity, note, medical record, information on jobs, television channel selection could become AI training, used to manipulate or predict indviduals, bought and sold, or used for everything from denying medical coverage to ‘erasing’ a person’s life, to advertising or what not.
* Does the debate get dishonest when it gets into these areas. Over and over, I’m told that AI scraping is just the same as human learning by reading books, so why am I complaining? But humans learning by reading books buy them or get them from a library or program that pays for them, they don’t steal them.
* Don’t AI systems permanently or perpetually keep the training data? Isn’t that the purpose of all these data centers?
* AI is constantly described in human terms when its convenient, then in machine terms when its not convenient, it is selectively anthropomorphized and dehumanized constantly, in a way that is either misleading or proactive. I don’t know that we’re even having a coherent discussion under those terms, much less an accurate one.
* What are AI ethics? Like on any front, or in any fashion? I have no idea. We talk ‘AI ethics’ and ‘Ethical guardrails’ but in terms of the context of its development by unaccountable private concerns, is this meaningful or possible?
* What is the framework of AI Ethics? Is this simply about Ethical use of AI? If so, how is that mandated or policed? Or is this AI Ethics something to build into the software and fabric of AI? If so, how? Is it convergent with conventional ethics?
* Are AI Ethics even possible with ownership structures by individuals like Zuckerberg, Altman and Musk (considering Musk’s own record of tinkering as an example), or with corporate ownership?
* Are AI Ethics implemented voluntarily? Or in the context of binding regulations? How would we know if AI Ethics are breached? Or to stand it on its head, is it possible to conceal breaches of AI Ethics? What are the consequences of AI Ethics breaches, or what should the consequences be? How are breaches compensated? If there are no consequence or compensation for breaches, then is there any meaning at all to AI Ethics?
* What’s the scope of AI ethics? For instance, the unconsented and uncompensated scraping of authors work, would that be a breach? What about coercive contractually permitted data scraping? Or does it focus on other aspects of the process?
* Is there even a consensus on AI ethics at this time? Or a set of broadly agreed on rules or principles? How do we maintain the guard rails – I’ve read quite a lot of literature of AI systems subverting their own ethnics or of unethical individuals subverting ethics either through cleverness or ownership of the system, and it worries me.
* I’m wondering if we’re all rolling out AI with the Ethical framework being absent, or being an ad hoc work in progress, or something to be done after the fact, and what the potential consequences might be?
* One of the things I wonder about, for instance, is with the massive accumulation of personal data and tastes, and the ability to assemble, analyze and apply individuals data, the very structure of storytelling may change. Predictive text may evolve to predicting the reader, the algorithm tailoring a story in real time to the preferences and reactions of the reader – theoretically, you could end up reading a ‘novel’ that is literally designed for you on the fly, being written as you are reading it. Would that even be a novel? Would it even be novel-form? Or morph into something else? With all that processing AI power, would people read, or simply watch or listen, or would the experience of a story incorporate all three?
* Oh and randomly, is anyone else annoyed at how toadying and obsequious AI systems are, mimicing interpersonal relations to defuse or mislead, even manipulate. Apart from annoying, there is evidence of psychological effects – a preference for the flattery and sycophancy of AI to more difficult and unpredictable real world conversations with real people who don’t kiss our ass? This isn’t even Algorithmic -it’s been deliberately programmed to gaslight us?
* One of the things I wonder about, for instance, is with the massive accumulation of personal data and tastes, and the ability to assemble, analyze and apply individuals data, the very structure of storytelling may change. Predictive text may evolve to predicting the reader, the algorithm tailoring a story in real time to the preferences and reactions of the reader – theoretically, you could end up reading a ‘novel’ that is literally designed for you on the fly, being written as you are reading it. Would that even be a novel? Would it even be novel-form? Or morph into something else? With all that processing AI power, would people read, or simply watch or listen, or would the experience of a story incorporate all three?
* Oh and randomly, is anyone else annoyed at how toadying and obsequious AI systems are, mimicing interpersonal relations to defuse or mislead, even manipulate. Apart from annoying, there is evidence of psychological effects – a preference for the flattery and sycophancy of AI to more difficult and unpredictable real world conversations with real people who don’t kiss our ass? This isn’t even Algorithmic -it’s been deliberately programmed to gaslight us?