{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreigb66argwzb6j2o5ayfovl2umoobwq6nv7rxxtmrcaawhekavpapy",
    "uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mitgq4tdapf2"
  },
  "path": "/t/ai-as-co-collaberator/174957#post_9",
  "publishedAt": "2026-04-06T10:42:53.000Z",
  "site": "https://discuss.huggingface.co",
  "tags": [
    "paulolden1/432-a-journey-beyond · Datasets at Hugging Face"
  ],
  "textContent": "This way of using AI is exactly how I used it while writing my novel “432 A Journey Beyond” (English version: “432 A Journey Beyond”) which you can find on Amazon but also here on Hugging Face as a public dataset paulolden1/432-a-journey-beyond · Datasets at Hugging Face ). I “collaborated” with 5 different AI chatbots while writing the novel, but none of them were used to draft sentences or entire chapters. They were used to brainstorm together, to look for references to places and technologies, to check the consistency of the narrative map, and finally as initial beta-readers. Initially, I was very excited and it seemed to work exceptionally well, but over time, I had to learn to be very, very careful about the tendency to complacency. AIs are programmed to often agree with you and to flatter you, essentially telling you that you are a genius! This can be very dangerous. I learned initial prompting to limit this problem, but I didn’t It’s a solution. I’ve also used a different account and a VPN to hide, pretending not to be the author. In some cases, it was really hard to force them to be objective: if the prompt, for example, simply says “don’t be complacent,” the model risks becoming overly critical, slamming everything to please you, assuming you want to dismantle everything. In any case, if you take this risk into account and that this tendency toward sycophancy needs to be monitored, I firmly believe that AI used as a collaborator is absolutely the best way to use AI today!",
  "title": "AI As Co- Collaberator"
}