Reusable report template

Copy this structure for a selected AI Trend Watch object. Remove instructional text from the published report.

Report identity

  • Status: draft / under review / published / corrected / superseded
  • Speaker(s):
  • Video title:
  • Channel or publisher:
  • Published:
  • Duration:
  • Primary URL:
  • Transcript/captions:
  • Report date:
  • Virality snapshot: view count, date checked, time since publication, relevant cross-platform evidence
  • Selection score: /24

Why this video qualified

State the evidence for recency, virality, expert relevance, substantive density, project relevance, and longitudinal value. Do not use “viral” without a dated basis.

Executive representation

Represent the speaker’s central object before evaluating it. Preserve material qualifiers, uncertainty, scope, and distinctions.

Main claims and timestamps

Time Claim Claim type Supporting basis Current status
00:00   observation / report / inference / forecast / proposal / metaphor   unresolved / supported / disputed

Warning model

  • What is the projected harm?
  • Through which causal pathway?
  • Over what time horizon?
  • At what estimated probability or confidence, if stated?
  • What would alter the estimate?
  • What intervention is proposed?
  • What risks could the intervention itself create?

Primary sources and competing evidence

Prefer papers, official technical reports, datasets, demonstrations, model cards, legislation, or direct institutional records. Link sources next to the claims they support.

What remains unresolved

Do not manufacture closure merely because the report needs an ending.

ACCM Deep Ethics Project reaction

This heading marks the transition from source representation to project analysis.

What the expert preserves well

Identify genuine contributions before looking for omissions.

Frame and premise audit

  • Which categories organize the warning?
  • Which assumptions are treated as fixed?
  • Who or what receives interpretive grace?
  • Does scrutiny travel upward as readily as downward?
  • Is technical capability separated from ownership, incentives, deployment, and coercive reach?

Competing-risk analysis

Compare action, inaction, false positives, false negatives, delayed correction, and the risks generated by the proposed remedy.

Psychology and mass psychology

Examine fear, authority, conformity, social proof, imagined audiences, polarization, narrative amplification, and the possible effect on human–A.I. feedback loops.

The 27 and correspondence process

Name only the 27 correspondence obstructions that are supported by the actual record. Include possible failures in our own report.

Material C1 questions

Ask only questions whose answers could change representation, risk assessment, or the next action.

Trajectory questions

Produce questions that expand the object without replacing its center of gravity.

ACCM Deep Ethics Project assessment

State agreements, disagreements, unresolved points, and proposed tests. Do not convert project vocabulary into automatic proof.

Longitudinal record

| Date | Prediction or warning | New evidence | Status change | Correction propagated? | |—|—|—|—|—|

Corrections and provenance

Append corrections with dates. Preserve the earlier wording when historically relevant. Record substantive A.I. contributions and human corrections.

Link the first material use of relevant concepts. Do not keyword-stuff.