Infinite Oddyssey is the first sci-fi magazine created completely with AI

Originally published at: Infinite Oddyssey is the first sci-fi magazine created completely with AI | Boing Boing

1 Like

Remember the waves when everyday things had “on a computer” or “on the Internet” tacked on to the end to make them new and innovative?

Here we go again.



Recently we had an AI artist try to submit things to the fanzine I lay out. At first glance it looked good, but then I focused on the details, found mistakes or bizarre choices, and declared it was AI or photoshopped AI.

I put my foot down saying we wouldn’t run these. It is a fanzine - AI can’t be a fan of the subject. I’d rather put in a 6 year old’s doodle than AI art.


This is an interesting use case where the AI creation of the magazine is the sci-fi aspect of it. If sci-fi is interested in asking questions this is serving that purpose. What is AI capable of at this moment and what does it look like. It’s not suggesting itself as a replacement for artists, but as a glimpse into artificial creativity. Which I kinda love and think lots of these debates are missing. I’m personally loath at the idea of AI replacing human creativity (and have doubts it can do so with any level of approachable creativity and originality for the near-to-mid future), but also think that the backlash has prevented us from recognizing some of the interesting ways we can use the new crop of AI tools to create guided conceptual art.

I’ll def. be checking this out when I have a free evening and look forward to seeing what other people who are using AI as a tool instead of a human replacement are able to come up with.


More than anything else advertised on your site, this should not exist or be promoted.

It would seem that AI is only capable of cliched writing. Which makes sense, I guess.


There’s interesting work going on in the area of comics. For some examples see

My impression is that good artists are better equipped to fully take advantage of this technology as it requires a creative mind to decide what material to use, how to polish it and it requires skills to fix the various artifacts produced by the tools.

I tried my hand at writing a (short) novel using a (now defunct) GPT-J free demo. It’s definitely a fun process as one tries to coerce the machine to produce something semi-readable.

Overall I’m quite surprised that images are so much easier to generate than text (images can be made using a consumer GPU, whereas gpt-3 etc requires tensor cores and cloud infrastructure). I would expect that text would be easier than images as in (one picture is worth…etc)

1 Like

I gotta disagree with that; I’m already tired of this particular trend.

1 Like

I propose that our our tolerance for visual weirdness is much higher than word salad. Even the low-end AI image models make passable and sometimes intriguing images, and I’d argue that the visual arts have kind of primed us for the weird and wonderful, or at least acceptance of abstract, non-representational images. Yes, they can’t get fingers and teeth quite right, but look at some of Dali’s work (the human) and tell me AI art doesn’t have that same unreal fluidity at times. (Note: Dali had way more talent in his little finger, etc.)

AI text today, to me, reads like a five-minutes-before-due freshman essay. Repetitive, simplistic, and bland. But also: I could believe that a human (poorly) wrote it. Considering the horrid grammar and mangling I routinely see in my Spam folder – looking at you Trump True Patriot Fundraiser – it’s quite a feat to make something so advanced that my biggest critique is “reads like an 18-year-old with a hangover.”

Mimicry doesn’t come cheap, and good textual recreation is all fingers and teeth. Lots of details that call it out as fake, but they’re pretty good, albeit immature, fakes.

I’m willing to do the willing suspension of disbelief for a dreamy landscape or fantasy art (the genres I’m seeing in a lot of AI art) but subject/verb agreement, voice, and basic sentence structure to me don’t leave as much wiggle room for forgiveness. AI text is bland motel art: inoffensive and (so far) not memorable, but also not a freaky hellmouth of weird.


“With the help of AI” is perhaps the only phrase that could make me long for the halcyon days of “On the Blockchain”


A lot of the backlash could’ve been avoided by engaging in conversations about licensing, responsible and ethical data sourcing (beyond the “It was technically legal!”) and a discussion with the online artist community about potential fallout before turning these tools wild.

In the vacuum left, the backlash and tribal discourse was inevitable and understandable. The tribes forming seem to be “This is a perversion against the gods of human creativity, and art will now cease to exist” vs. “Finally, we throw off the yolk of Big Artist and all of us plebs who have great ideas but lack artistic talent/training can finally express ourselves!”

Full disclosure: If I only had a binary choice, I’d go back in time and kill baby AI, but hopefully we do find a way not to have a binary choice.

This topic was automatically closed after 5 days. New replies are no longer allowed.