Delighted with this good neutral coverage in Business Insider by Beatrice Nolan, regarding my now 99 books generated with help from AI. The author did a good job contextualizing my efforts within the broader scope of what is happening with AI books.
The Octave of Time is book #99 in the AI Lore books series, by Lost Books, Canadian AI publisher.
The text was is all human-generated, and was written chronologically as a sequel to The Lost Direction. However, due to circumstances, the book was unfinished, with only maybe a third or half of the full tale being told here, of the settling of the island of Ovarion, after the collapse of Quatria back into the Hypogeum.
This book features a returning cast that includes Tob Gobble, among many others. It also references quite a lot the House of Life (though the volume titled that was only written much later).
All the images are done in Midjourney. This was interesting for example with depicting recurring characters like Tob Gobble, who takes on many new and surprising faces within this volume. And MJ had a completely terrible time trying to depict the half horse/half rabbit creatures known as goleks. (There is one of these books that has some awesome pics of goleks, but I can’t remember offhand which one it is anymore!)
Here’s the preview of the interior art:
One thing I love about doing images in this style is that because of the “dvd screengrab” style I use on many of them, I’m left with the impression of there being a movie/movies here that “must have” existed, and the memories of which are nearly at the tip of my tongue… and that’s an incredible feeling; plus this feeling of being able to like “open rooms into my subconscious” and let others peer inside.
By the way, here is the original reference to the concept of the Octaves of Time, as described originally in the Lost Direction.
When Benda upset this natural way of things by mortally wounding Jan Re, he incurred both a blood debt to support Jan Re’s family, but also a ritual debt, for having changed the pitch of the tones which make up the Octave of Time. The cycles performed in Quatrian Society under the priests of the Hypergeic Temple Mount had as their intended function the harmonization of these Octaves in the present moment with the mythic and historical root tones (which were one in the same in this society).
This was not, to those people, idle speculation, or merely a symbolic conception of time as cyclical. This was, in a society still wedded to the Hypogeic powers, a concrete experience of how the present could impact the past. Thus atonement was also attunement, a continual re-tuning of the very stuff of existence.
I’m kind of an asshole about privacy. Which is why when GDPR came out, I was all over it. It’s not perfect by any means as regulation (let alone enforcement), but it’s a strong step in the right direction.
Which is why it annoys me so much when companies don’t follow it. Even though I’m not an EU resident/citizen, I’m a stickler for it, because for the most part, the principles enshrined in it also just make good product sense.
One of the articles I often come back to is Art. 25, Data Protect by Design & By Default, one of whose clauses reads:
“In particular, such measures shall ensure that by default personal data are not made accessible without the individual’s intervention to an indefinite number of natural persons.”
For context: Midjourney has presently three paid tiers, at $10, $30, and $60 per month. It is only at the highest paid tier ($60/mo), that Midjourney offers what they call “Stealth image generation.” Unless you pay $60/mo, and also actively turn on ‘stealth mode,’ all images that you create using the service will be automatically made public in their member galleries (accessible to an indefinite number of persons), and in a section of their site where users rank pairs of images to help train their AI models.
Also, it’s worth noting that the default generation behavior – if you’re using the public Discord (I use my own private server, partly for this reason), all your images are shared in a terrible mess with everyone else’s generations. I’ve seen rationale from the CEO claiming that the reason for this is acceleration that occurs, synergy between many people’s prompts, etc. I guess that’s fine, but I think it breaks GDPR.
From the FAQ on Midjourney’s logged in account page:
“We are building an open-by-default community focused on collective exploration and fun. If you have a need to opt-out of this and be private-by-default you can subscribe to the $60/mo pro plan and activate stealth generation with the
/stealthcommand.”
As a privacy professional with certifications in GDPR compliance, I find this pretty abhorrent as a practice. If privacy is indeed a human right (I believe it is), then it is reprehensible to only offer it for sale to those willing and able to pay the highest price for it.
Hence, I took my frustration on this matter to multiple Data Protection Authorities in the EU, after receiving no response from Midjourney staff about any of this over several months.
Two issues: as a non-EU person, I don’t really have standing to file a complaint insofar as it impacts me personally. So I filed it as a “request to investigate,” and gave them all the necessary information.
Second, it seems like the DPAs are extremely slow, bureaucratic, and ineffective as enforcement bodies. I could be wrong, but this isn’t my first rodeo, and I’ve seen how brush-offs work in the past. I sent my complaint out to email addresses I found for about 7 or so DPAs in different countries, and nearly half of them were bounced. I noticed also France’s CNIL doesn’t even have an email address… wtf?
Anyway, if you’re at all concerned about these issues, and not into letting tech companies get a free pass because they are Americans and think global laws don’t apply to them (or ought not), AND you’re an EU resident or citizen, I would strongly encourage you to find the DPA in your country and file complaints against Midjourney so that they – and all the others – end this gross practice of selling privacy.
Just wanted to post finally this Midjourney image set I did of ballet dancers underwater and/or “wrapped in plastic.”
Here’s the full set.
Here’s a sample image:
These images are a secondary offshoot that arose while making the book The Plastic Prison, which itself is one of my favorite most evocative books I’ve done. Wrapping people plastic and putting them underwater, etc. is one of the amazing things you can do with Midjourney: putting them into situations that would be deadly or difficult/impossible to photograph safely.
These images just blow my mind. Here’s one more for the road (btw the full set has 99 images).
Learning how media coverage works over the past few years, and especially lately has been… interesting. Authors do not have an easy road of promoting their works. Literally every person on planet earth has written a fantasy novel, and it’s basically impossible to get press about it. Unless you’re already known with a well-developed following.
So I did what anybody would do in a situation like that, I used generative AI to write 100 books, and suddenly (I won’t give away all my secrets) people started to notice. Combined with other elements, like viral gen AI images on social media, and an expertise in content moderation and disinformation, the endeavor started to seem newsworthy. But it was a long uphill slog getting there.
Eventually, you reach a point where the fact that you got coverage is the thing that becomes newsworthy. I got onto Newsweek because I got onto CNN. I got onto Futurism because I got onto Newsweek, and the circle goes on and on.
So where to next? I realize there is still a story central to this whole operation which has not yet been adequately told. My story. Not just the punchline or the gimmick or whatever, but the human path that brought me here.
It all starts with content moderation; perhaps the best coverage of that actually happened quite some time ago, via David Farrier’s long-ish interview with me, archived here. But that was in a different era, before this current batch of tools hit, and before AI really exploded onto the scene…
Anyway, in an effort to have a chance to further tell that story in the media, I put out this press release a few days ago, and I’m planning to go into greater depths about the roots of my work in the… shall we say, terrors… of the world of content moderation, and how it took me a long time – maybe three years (?) to nearly or almost fully recover from its effects. I’ll of course in the interim use this blog as a sounding board as I figure out how best to tell that final story, the story of me.
Sometimes when undertaking art projects that are very big or take a long time, it can be useful to take up a new artist moniker specific to that project. Something to dump all the psychological contents of the process into, a container, an entity which can become itself an egregore, an invocation. And which, when you’re finished, you can leave behind.
I had such a name like that come to me the other day during meditation. That’s a topic I intend to write about it in its own right in more detail (meditation is its own kind of hyperreality), but suffice it to say sometimes in the meditative process, things seem to arise all on their own, through some deeply intuitive process which is for the most part opaque.
The name that popped into my mind was ‘Baofa,’ which I sat weighing in contemplation for some time, and it stuck with me. I thought it had kind of a cool, singular quality to it, like “Madonna.”
On ending the session, I started querying different sources online to see if it was really a thing, and found that apparently this word has a meaning in Mandarin Chinese, and its characters are 暴发. Google Translate seems to confirm the translation there… Though the primary translation appears to be “outbreak,” the secondary ones listed there are: get rich quick, suddenly become important, suddenly become rich.
I didn’t undertake my generative AI art projects in order to become “rich,” but given the headline Newsweek wanted me to write an article for, and other comments received in response to that piece, I think it’s only too funny and ironically fitting in the end I guess to adopt, essentially, “get rich quick” as a name.
I don’t know Mandarin Chinese, nor have any connection to that culture personally. I don’t know if somehow my subconscious mind tapped into some store of universal human knowledge to taunt me with silly name-calling, but I guess I don’t necessarily rule it out either.
Interestingly, I’ve been looking around for comprehensive lists or even references to generative AI artists by name, and they are hard to come by. Who are the artists working in this space? By what names are they going & why?
There are a great many mysteries in the world of promoting your work as a writer. I just wanted to share some insights around traffic and sales to my Gumroad, following coverage of my work in Newsweek.
Here’s a graph for Gumroad for my last thirty days:
You can see at right a spike in traffic (grey bars) for May 15th, when the article dropped. Page views for the store peaked at 424 visits the day after, as it did sales. During the period following the piece, I sold a total of 22 books, which I am confident attributing to the Newsweek piece, and its ripple effects.
That is not a huge batch of sales, but I’m pretty happy about it regardless. You can see though, how I still get somewhat consistent low sales prior to that. In the weeks prior to that during the past 30 days, I still made 13 sales, and one batch of those was someone buying a bunch of books at once.
Now, it’s possible if my original article in Newsweek was different – either different contents, or a different author covering my story, that those figures would be different. It’s difficult to say.
In any event, I just wanted to transparently share this sales data for other authors wondering just how much of a windfall this kind of coverage really is. The short answer is, not huge in terms of sales, but quite big personally for the doors it opens for me to get further media coverage.
I saw this thread on the AbsoluteWrite.com forum discussing my Newsweek article, and thought it was worth bookmarking here, along with capturing both some good points and some items worth responding to.
As forum threads so, it kind of wobbles all over the map, before apparently getting shut down by moderators for generating too much controversy.
One user there, Schaun, has a lot of strong arguments that I mostly agree with. He proposes there is one catch in my plan though, and foresees this as one of several outcomes:
Most other people can easily replicate his approach, in which case he’s going to see his market flooded and will have to put more and more time and effort into making less and less money, and may see the bottom drop out entirely once his very niche audience diminishes to only those who weren’t in it for the novelty effect alone.
I agree that anyone can do what I’m doing. The AI tools are accessible to all. I think however there are a few things I’m doing well, and a few other things that are somewhat unique in my approach.
Things I’m doing well:
Things I’m doing that are somewhat unique:
There’s probably more to add to both parts of that list, but that is enough of a trailhead for right now.
All of that is to say, I recognize that I have no real “moat” apart from the speed of production, the quality of my products, and the volume of items already on the market place. Anyone else can (and will) come along and do more, faster. Of that I have no doubt. But I have little fear that they will compete in my space, because the space I’ve carved out narratively is very well colonized. It’s a brand, a niche, and an established fan base. If people want my stuff, they will come to my channels to get it.
Even if other people start adopting a similar approach, that’s fine. It won’t impact the time it takes me to make the books, nor will it impact the amount of enjoyment I get from the process. My sales are small, but steady. So there’s not really any big cliff for them to drop off from.
Incidentally, later on someone in the thread named Chase proposes, regarding the bottom dropping out, “He’ll just make someone in the Philippines or Columbia or Nigeria do it for him.” This wouldn’t make any sense for my situation, because the process is a creative one that I love. If I farmed it out to someone else, it wouldn’t be fun anymore, and they wouldn’t be able to produce the quality I am able to produce.
Lastly, I don’t think there are probably any buyers in it for “the novelty effect alone,” because I get the impression most buyers didn’t come to my books because of my incorporation of AI technology. They came because the stories sound cool, and they come back because of the consistent quality of the product.
Moving on to other topics…
I see a lot of probably wrong information on copyright in this thread. It’s not worth trying to correct all of it, but a few items do jump out at me.
Hickson writes:
I see it as an ethical deficit on the part of those who decided to scrape the internet without seeking permission.
I understand why people feel this way, but US Appeals Court in wherever decided that scraping the web is legal. I think if people want to more meaningfully engage in the “ethics” side of that (apart from legality), then it needs to be made more clear which code of ethics ought to apply.
Personally, I voluntarily try to follow more or less the AI ethical guidelines of the Alliance for Independent Authors – even though I am not a member. Their primary requirement simply seems to be disclosing the use of AI, which I do both within the books, and at the publisher/account level.
Again regarding copyright, user Helix writes:
It’s closer to sampling IMO in the sense of taking elements of other people’s work and using it elsewhere.
Here are my thoughts as to why is actually not like sampling at all. To recap them in brief, sampling makes use of whole cloth excerpts from the works of someone else. Generative AI does not. It instead measures attributes (dimensions) of data and relationships from many works in aggregate, and then uses that to develop entirely new works. There is no use of any other work’s specific contents that are identifiable in any of my books.
User CMBright states:
It has been mentioned repeatedly in other threads here that no one owns the rights to AI produced work.
This is wrong. My understanding is that the US Copyright Office, which is not a law-making body, made a statement that a single AI-generated image cannot be copyrighted (in its non-binding opinion), because it has no human author. I think even that is a probably flawed and wrong opinion which will certainly be legally challenged in the near term. They did, however, I believe state that the entire contents of the AI comic book at issue could be copyrighted, because of the unique arrangement of parts.
More importantly, my books are not copyrighted in the US. They are copyrighted in Canada. (Canada also has different rules around Fair Use btw, which is called here Fair Dealing.) People in the United States often mistakenly believe their country is the only one, and that its laws apply everywhere.
User Brigid writes:
I wonder when whatever program he used will be knocking on his door looking for a cut of the proceeds, seeing that it was co-written by whatever program.
In general, my understanding is that most of the services use to generate AI content do not claim ownership or exclusive rights over the contents generated. It varies a little from service to service how it is worded, but by and large their licenses permit these kinds of uses. So no, I don’t think this will be the case.
Dipping back out of the copyright questions again, a couple more bits I wanted to capture from Schaun’s replies. First, I think the comparison to synthesizer music is spot on:
I was saying the guy was producing something. He used a tool to do it. I don’t personally have a taste for the kind of thing he produces, but I’m not going to say it was wrong for him to produce it. I’m not saying synthesizer or any other music isn’t real music. I’m saying it is real music, even if I don’t personally prefer it. I’m saying the books the guy wrote are real books – not copying or stealing from other people – even if they aren’t books that I would personally enjoy reading.
Many people in that thread are bent out of shape that the ebooks are only 2-5k words in length. Not a single commenter anywhere or in the coverage I got, however, mentions the large image sets which go into these books to add value. Not to be overlooked from a marketing and from a sheer “fun-ness” perspective.
Schaun also brings up an excellent point here, regarding the better quality AI works we see popping up across the web:
Those remarkably good examples aren’t the product of just throwing a prompt into an AI. They’re the product of hours and hours – and probably days and days – of human intervention, reworking, and hand-editing. […]
When we see amazing results, we shouldn’t say “wow, a computer did that. ” We should say “wow, a computer plus an insane amount of human micromanagement and prompt massaging did that.”
Anyway, that’s enough beating of this particular horse. I’m happy to see people talking about the work, but I’d much prefer they talk about the actual contents & structure of the work, because to me that is about 1000x more interesting than all these secondary issues around the technology. But if this is the doorway to get people to talk about the work at all, then okay!
Been thinking more about Stephen Marche’s piece on generative AI. He compares it to Hip-Hop, in certain respects. The most salient quote from it is here (source):
To make hip-hop, you don’t need to know how to play the drums, but you do need to be able to reference the entire history of beats and hooks. Every producer becomes an archive; the greater their knowledge and the more coherent their understanding, the better the resulting work. The creator of meaningful literary AI art will be, in effect, a literary curator.
I think he’s basically right, but there’s a foreseeable rhetorical danger here in comparing generative AI to Hip-Hop, because to many people Hip-Hop is synonymous with sampling: taking a beat or a hook or whatever wholesale out from another song, and re-purposing it into a new thing.
If you say gen AI is like sampling, then that opens up potential criticisms around copyright, that are – I think – incorrect with how the technology actually functions.
My somewhat non-technical understanding of how gen AI works is this: generative AI tools do not take “samples” or clips from other pieces and collage them together. Like an arm from here, an eye from here, lips from another source, etc. That’s just not how it works.
How it works is that many sources are analyzed for their dimensionality (being here something like attributes or characteristics). This piece is red. That piece depicts a human, this one a sky, this one a puppy peeing on a fire hydrant. From those aggregate measurements of dimensionality, an entirely new thing is invoked by user prompts that in no way incorporates a “sample” of the original pieces it was trained on.
Okay, then there’s the question of whether or not creators agreed to have their works “measured for dimensionality,” but in the United States, that seems to be a moot point, because on appeal, web scraping was deemed to be legal. So if you want to argue whether scraping & then measuring what you scraped is “ethical,” the court seems here to have some guidance, though certainly there are additional dimensions to the conversation that are worth interrogating.
Adobe has taken the approach of only using source images it owns rights to, and I think that’s a good idea for their particular usage (though I think Firefly image quality is pretty shitty).
Anyway, more to say on this, but other things press for the moment.
Futurism ran a response to my Newsweek article about my AI lore books. Notably, the title of the Futurism piece is: “Man Uses AI to Write 97 Terrible Books, Sells $2,000 Worth.“
It’s fine if that author didn’t like the 1 out of 97 (now 98) books that they gave a “quick read.” But I think it’s wrong to say that they all suck, because they didn’t like the one that they skimmed. It’s the equivalent of reading one or two paragraphs out of a 1,000 page book and deciding the whole thing is bad.
In actual fact, the volume they didn’t like, Inside the Hypogeum, is actually one of my favorites. Along the way, with the many discoveries I have made, that book was a turning point, because it proved that I could indeed effectively use AI to write good quality lore that fits into the canon of my universe quite neatly (never mind all the awesome art in that book and all the others). And then it is of course very dear, because of the centrality of the Hypogeum to the greater Quatrian mythos. If you don’t already Helmoquinth, Anthuor! though, I can see that the whole thing is probably confusing and strange.
They also included in the Futurism piece a predictable criticism of that book as being, “a meandering explainer of the fictional locale and legends — but no discernible plot or developed characters to sustain reader interest.” Obviously, they missed the bullet point that this is world-building; that’s literally how this genre works. It is, as I like to say, lorecore.
Leveling this criticism at literally any of these nearly 100 AI Lore books is like saying “Appendix F of the Lord of the Rings is really boring.” And you know what, without any other context, or reading the work it comments on and fills out (in my case, that would be The Lost Direction), Appendix F on its own may very well be boring. But when you take it as it was meant to be taken, among all the other voluminous writings of Tolkien, it takes on a numinous quality. (And actually, when I look it up, LOTR Appendix F really is awesome!)
Realistically, I guess you could say that my main innovation here is that all my books are essentially appendices of one another. And that is by design. It is not for everybody, but for those people who like it, they REALLY like it.
Originally, in the Futurism article’s first published version, they included the blatantly wrong line:
Boucher claims he’s had repeat readers who loved his books, but there’s little evidence of that.
I wrote to them asking for a correction, and showed them proof of sales, with buyer identities redacted.
Without responding to my email, they since pulled that line and added the cryptic qualifier: “Updated to remove speculation about the existence of repeat buyers.“
For future reference, I eyeballed a sales spreadsheet and came up with this break down of repeat buyers.
My math might be wrong, but my back of the envelope calculation then suggests that about 8-9% of buyers bought multiple books. And those repeat buyers comprise about 40% of my total unit sales. Amazing!
That’s a pretty significant proportion of all my buyers being repeat buyers who know what the content is, know what they’re getting into, and come back for more again and again – because they love it! All with no advertising, no social media promotion, and no overhead…
The thing is though, Futurism never reached out to me for comment or additional information before or after publishing. I could have easily given them this information.
I wasn’t able to find any policies regarding editorial ethics on Futurism’s site – so I can’t speak to what standards they believe themselves to be bound by – but I do happen to know that the Society of Professional Journalists Code of Ethics does include a variation on the well-established journalistic practice of a right of reply. SPJ’s code states that ethical reporting should:
Diligently seek subjects of news coverage to allow them to respond to criticism or allegations of wrongdoing.
I have no idea what Futurism’s official policy is, but I can say that they did not follow this guideline in the case of their reporting on my books.
(Nerd Tangent Incoming: This notion of a right of reply seems to have roots in Roman law, by way of the Latin legal maxim, Audi alteram partem – “let the other side be heard as well.”)
Anyway, putting all that aside, and putting aside the fact that they didn’t like the 1 out of 100 books they looked at (to each their own), the criticism contained in the piece seems to follow the common shape I saw in comments elsewhere on the Newsweek post:
But none of those actually hold up to scrutiny.
If it’s so easy, why doesn’t everyone have hundreds of AI books they published? Hint: because it’s not actually that easy to get consistent results of a decent quality. Doing what I’ve done took a hell of a lot of work, and a hell of a lot of trial, error, and discovery along the way.
If the quality is so bad, why did one person alone buy 39 different titles, and dozens of other people bought multiple copies?
If AI writing is so easy and so bad, wouldn’t it more logically follow that I shouldn’t actually be able to make any money off it at all? Why then does the implied or expess expectation of this kind of commentary always seem to point to some notion that I should actually be making more off this?
Shaming me for not making enough money in the short term is lame. Especially since this is a long game. This is literally long tail book sales. One person in comments somewhere suggested with the time I’ve put in, that I’m only making somewhere in the ballpark of $3 an hour (btw this is not my full time job). But that ignores how these products will compound over time. I can potentially sell these for the rest of my life, and beyond. What will my sales figures look like after I reach 500 interconnected books – or 1K – and how big will this get if I keep getting MSM press coverage (even occasional bad coverage like this)?
It’s also funny that Futurism chose to conclude their article by mocking my good reviews on a book I wrote prior to using AI (but the subject of which is entirely about AI, and how it will control our lives), Conspiratopia. I’m not sure what they think that proves. If they did their homework, then they’d see that it got even better reviews on Goodreads!
In the end, my sales figures seem to question the validity of the assumption that purely human-generated content is somehow better or more preferable to audiences. Can it really be worse than a human author who hasn’t done their diligence in composing an article?
Also, the subtitle of their article, “Human writers probably shouldn’t be too concerned… yet,” misses something important: I am an early mover in this space. Yeah, my moves are many and imperfect – but I don’t hide that. While other people are still busy debating the validity of using the technology, I have generated 100 books with it, gained a ton of experience, developed a dedicated fan base & sales channel that I control with little outside interference. As the progress of the tech continues to explode, this is going to put me extremely far ahead of the pack.
Oh, and by the way, just for fun, I ran Futurism’s article through ZeroGPT, a supposed AI-content checker, and it suggested their piece was 12.87% AI written. True or false? Unfortunately, there’s no way to be sure.
Powered by WordPress & Theme by Anders Norén