For a moment I entertained the idea that these are intentionally bad to get people to buy them as gag gifts. My kids and I certainly had a good laugh looking at the pictures in blog. That second picture of the jaw sticking out had my son ROFL-ing.
I've been wondering what the long term result on peoples perceptions of reality will be after all this AI slop. I've noticed a lot of the times I can spot AI slop videos because they just don't match what I know to be true, I can think "That's an AI video of a fox because I know foxes don't act/move like that" But then the only reason I know that is because I've seen hundreds of videos on the internet before AI generated video was a thing. But someone who grew up seeing AI slop from the start doesn't have that firm grasp on reality to spot fake content from.
The concept of consuming AI generated content for children has always baffled me.
We collectively have a virtually infinite collection of already existing hand crafted quality content filtered over the years in the form of children stories and tales that we can pick and chose from to read to our children. We love telling stories especially to our children.
Why would ANYONE be enticed by the idea of using AI to generate tales when there are so many out there to tap from is really beyond my comprehension.
The only way this ends well is that we end up with extreme regulations in all places to enforce ownership, from software (because of insecure code due to excessive vibe coding) to children's books (because we end up teaching them crap).
Totalitarian as hell, but I don't see any other way.
Just like Vibe coding, I think this article is trying to convey that, when children’s books important not to rely solely on AI, but to consciously strive to create high-quality works.
This just seems like laziness vs AI = bad. It's not like publishers are putting out masterpieces with human writing. They're cranking out minimum viable content as well.
I've found that by putting meaningful effort into AI storytelling, I can create bespoke stories that my kids love night after night.
My workflow is below: Caveat that it costs about $0.25-$0.50 to weave a book like this with Claude Sonnet and Gemini Nano Banana Pro. But to me the cost is worth it for the quality.
- Use Claude structured output and ask for page1, page2 ... pageN instead of an array of pages or wall of text.
- Pass a story arc as a set of values to the prompt. I.e. say each page has an emotional beat between 0.0 and 1.0. For a "man in hole" type of story: page1 starts at 0.6, page2 = 0.5, page5 = 0.25, page10 = 0.85. This ensures page 5 lands the "crisis" and page10 resolves higher than the start.
- For illustrations, have Claude generate the story text and an illustration prompt per page. i.e. page1: { "text": "...", "illustration": "..." }.
- For art consistency, add an "Art Direction" key to the structured output. Feed this into Gemini/OpenAI and ask for an art board visual guide & character reference sheet.
- Feed the page text, illustration prompt, and the art board to Gemini/ChatGPT images. I'm constantly surprised at the quality of the output.
Here's an example set of pages from a magic school bus style story about the immune system
It's not laziness. It's a shift of burden (bullshit checks) to the consumer. That burden is not trivial. It takes time and energy, so it's much better for the society when people don't have to do it often. LLMs (will) increase that burden by orders of magnitude as we're going to experience bullshittification of pretty much everything - entertainment, services, software, gadgets, science & research, healthcare, lawmaking, etc.
>Last week, I posted a visual demonstration of the sameness of AI-generated content. This makes the output easy to spot even if all the individual pieces are perfect facsimiles of what a human could create:
I think a point missing is that this output all looks the same because the prompters are not specifying much more than the barest minimum to get what they want. If you just prompt "generate a cover for my book 100,000 whys which is childrens book that answers their questions about science" then you get images like from TFA using the models default style. However, the models are capable of reproducing any great artists style and any content you want.
If you have seen the prompts for images on communities of enthusiasts you may notice that they can be quite long and specify considerable detail about both the content and the style of the output.
Here is one of the four above the fold on the front page of CivitAI for me right now, it has both a positive and negative prompt. Not that long because this is a fairly simple image. However the image doesn't look like the slop in the 100,000 Why's book covers or the many commercial signs and advertisements I'm seeing when I leave the house.
21 comments
[ 5.7 ms ] story [ 48.5 ms ] threadAI slop is just a more complete reimplementation of the "shovelware" from the 90s
https://en.wikipedia.org/wiki/Shovelware
We collectively have a virtually infinite collection of already existing hand crafted quality content filtered over the years in the form of children stories and tales that we can pick and chose from to read to our children. We love telling stories especially to our children.
Why would ANYONE be enticed by the idea of using AI to generate tales when there are so many out there to tap from is really beyond my comprehension.
I'll start: John Rocco, How We Got to the Moon. (http://www.howwegottothemoon.com/)
Totalitarian as hell, but I don't see any other way.
Afaik, parents are super protective of their children and would never do something that could inhibit a childs learning
I've found that by putting meaningful effort into AI storytelling, I can create bespoke stories that my kids love night after night.
My workflow is below: Caveat that it costs about $0.25-$0.50 to weave a book like this with Claude Sonnet and Gemini Nano Banana Pro. But to me the cost is worth it for the quality.
- Use Claude structured output and ask for page1, page2 ... pageN instead of an array of pages or wall of text.
- Pass a story arc as a set of values to the prompt. I.e. say each page has an emotional beat between 0.0 and 1.0. For a "man in hole" type of story: page1 starts at 0.6, page2 = 0.5, page5 = 0.25, page10 = 0.85. This ensures page 5 lands the "crisis" and page10 resolves higher than the start.
- For illustrations, have Claude generate the story text and an illustration prompt per page. i.e. page1: { "text": "...", "illustration": "..." }.
- For art consistency, add an "Art Direction" key to the structured output. Feed this into Gemini/OpenAI and ask for an art board visual guide & character reference sheet.
- Feed the page text, illustration prompt, and the art board to Gemini/ChatGPT images. I'm constantly surprised at the quality of the output.
Here's an example set of pages from a magic school bus style story about the immune system
[image] https://media.discordapp.net/attachments/839188039229112353/...
I think a point missing is that this output all looks the same because the prompters are not specifying much more than the barest minimum to get what they want. If you just prompt "generate a cover for my book 100,000 whys which is childrens book that answers their questions about science" then you get images like from TFA using the models default style. However, the models are capable of reproducing any great artists style and any content you want.
If you have seen the prompts for images on communities of enthusiasts you may notice that they can be quite long and specify considerable detail about both the content and the style of the output.
Here is one of the four above the fold on the front page of CivitAI for me right now, it has both a positive and negative prompt. Not that long because this is a fairly simple image. However the image doesn't look like the slop in the 100,000 Why's book covers or the many commercial signs and advertisements I'm seeing when I leave the house.
https://civitai.com/images/134444826