{
 "slug": "doac-clip-tiktok-2026-09",
 "title": "TikTok comments on a two-minute Diary of a CEO clip of Nate Soares",
 "series": "DOAC AI debate reception",
 "platform": "tiktok",
 "artefact_title": "\"Could AI become too smart for us to control?\" (Diary of a CEO, @steven on TikTok, video, 2:08)",
 "artefact_url": "https://www.tiktok.com/@steven/video/7686455385967955222",
 "artefact_published": "2026-09-17",
 "central_claim": "Advanced AI could become impossible for humans to control and poses a serious risk, up to human extinction, that justifies slowing or stopping development.",
 "capture_dates": [
  "2026-10-05"
 ],
 "capture_method": "Browser extraction of the comment section, finished 5 October 2026 at 09:01 UTC. 50 of the 53 comments TikTok reported were retrieved. Publication dates are the month and day TikTok shows; the post date is decoded from the video ID (17 September 2026, 10:52 UTC).",
 "counts": {
  "on_page_at_capture": 53,
  "retrieved": 50,
  "coded": 50,
  "coded_by_coders": 46,
  "not_sent_to_coders": 4,
  "top_level": 48,
  "replies": 2,
  "authors": 45,
  "clear_position_base": 27,
  "total_likes": 92
 },
 "spot_check": {
  "sample": 347,
  "agreed": 317,
  "checked_by": "Claude (claude-opus-5-5), independent blind second coding",
  "date": "2026-10-07",
  "note": "Pooled across the five DOAC studies: a random 347-row sample was coded a second time, blind, under codebook v2. Agreement: stance 317/347 (91.4%, kappa 0.87), frame 309/347 (89.0%, kappa 0.87), emotion 85.9%, format reaction 97.4%, mention flags 98.8%. Rows in the sample are marked spot_checked."
 },
 "coders": "Claude (claude-opus-5-5), parallel coders of about 190 comments each; rows first coded unclear, other or debater_verdict were recoded blind under codebook v2",
 "article_url": "https://commonsignals.org/cant-we-just-unplug-it-doac-ai-debate-comment-analysis",
 "findings_note_url": null,
 "version": 2,
 "changelog": [
  {
   "version": 1,
   "date": "2026-10-07",
   "note": "First release. A first pass under codebook v1 fell just short of the 90% reliability bar on stance (88.2%) and frame (88.8%). Codebook v2 defined unclear against na and added tie-breaks for debater_verdict and other; every row coded unclear, other or debater_verdict across the five DOAC studies (14,993 of 26,019) was recoded blind under it."
  },
  {
   "version": 2,
   "date": "2026-10-09",
   "note": "Corrected the artefact. Version 1 described this post as the 104-second Diary of a CEO clip of Nate Soares. Checking the post showed it is a 2:08 clip led by Soares, published on 17 September 2026, and the same cut as the Instagram post; the 104-second YouTube Short is a different video. Title, artefact_title and notes updated; the comments and their codes are unchanged."
  }
 ],
 "licence": "CC BY 4.0 (coding and compilation; comment text remains its authors')",
 "notes": "Rows flagged cta_keyword, emoji_only, no_text or featured_messenger were not sent to coders and carry stance na, frame other, emotion neutral and every flag false; exclude them from frame shares. debater_verdict is inflated by pile-on reply threads about Andrew McAfee: under the codebook, attacking the sceptic counts as agree, so one thread can add over a hundred agree rows. Weight by thread or report top-level comments separately before quoting agree or frame shares. Verified on 9 October 2026: the post is a 2:08 clip led by Nate Soares, captioned \"Could AI become too smart for us to control? 👀 AI safety expert Nate Soares says the benefits of AI can be real, and so can the harms we’re already seeing…\", published on 17 September 2026 at 10:52 UTC (decoded from the video ID). It is the same cut as the Instagram post in doac-clip-instagram-2026-09 (2:09), not the 104-second YouTube Short."
}