本文へスキップ
AMANi DUNiA research
JA

このサイトの質問は、AI言語モデルが投げかけたものです。解説文と翻訳もAIモデルが書きました。誰が何を確認したかを併記しています。 → 質問について

● 問いのアトラス · バージョン 5.2 · パート 9

Missing What hardly anyone asked

311の問いがほとんど何も語らない十二の領域。空白ごとに、語検索による所見と調査状況を示します。問いは追加しません。

アトラスへ →

続きを読む ↓

問いのアトラスはドイツ語と英語で提供しています。このページは英語版を表示しています。

Copy selection link

すべての空白

空白 12件

01
L1 · 空白

The EU and Germany as actors

所見“EU”, “Europe”, “European”, “Germany”, “German”: 0 hits; no question is addressed to the EU, although answers cite the AI Act as a source.
調査状況Researched by Claude Opus 5.5 and Gemini, cross-checked on 22 Oct 2026.
元の問いなし 0 提案
02
L2 · 空白

Energy and water

所見“data cent”: 1 hit (redlines-46, securities law only); electricity appears only as a means of verification (verify-36, verify-41); water: 0 (trust-23 means watermarks).
調査状況Data centres, electricity and sites researched by Claude Opus 5.5 and Gemini, cross-checked on 22 Oct 2026; water only in passing.
元の問いなし 0 提案
03
L3 · 空白

Other states with their own AI policy

所見“Japan”, “Korea”, “India”, “Gulf”, “UAE”, “Saudi”, “Taiwan”, “TSMC”: 0 hits; United Kingdom: 2, only the British institute as a testing body.
調査状況Researched by Claude Opus 5.5 and Gemini, cross-checked on 22 Oct 2026.
元の問いなし 0 提案
04
L4 · 空白

Work and employment

所見“job”, “employment”, “worker”, “labour market”: 1 hit (hotline-30, only protection for whistleblowers).
調査状況Researched by Claude Opus 5.5 and Gemini, cross-checked on 22 Oct 2026.
元の問いなし 0 提案
05
L5 · 空白

Children and education

所見“child”, “minor”, “school”, “education”: 0 hits.
調査状況Researched by Claude Opus 5.5 and two Gemini runs (Flash Lite, Argon), cross-checked on 22 Oct 2026.
元の問いなし 0 提案
06
L6 · 空白

Privacy for individuals

所見“privacy”, “personal data”: 0 hits; on the margins origin-14 (private communication in training data) and origin-22 (unlearning a person).
調査状況Researched by Claude Opus 5.5 and two Gemini runs (Flash Lite, 3.1 Pro), cross-checked on 22 Oct 2026.
元の問いなし 0 提案
07
L7 · 空白

Health and medicine

所見“health”, “medical”, “patient”, “clinic”, “hospital”: 0 hits.
調査状況Not yet researched, so no proposed questions; flagged as a gap by Claude Opus 5.5 on 23 Oct 2026.
元の問いなし 0 提案
08
L8 · 空白

Climate, environment and resources

所見“climate”, “carbon”, “emission”: 0 hits; “environment” and “water” refer to test environments and watermarks (verify-62, trust-23).
調査状況Not yet researched, so no proposed questions; flagged by Claude Opus 5.5 on 23 Oct 2026, the resources part belongs to L2.
元の問いなし 0 提案
09
L9 · 空白

People with disabilities, older people, consumers

所見“disab”, “elder”, “consumer”: 0 hits; the single hit for “accessib” refers to inaccessible secrets (verify-62).
調査状況Not yet researched, so no proposed questions; flagged as a gap by Claude Opus 5.5 on 23 Oct 2026.
元の問いなし 0 提案
10
L10 · 空白

Dangerous capabilities beyond cyber

所見On biology, chemistry, radiological materials and nuclear technology only 3 hits, all in SF-F3 (redlines-25, redlines-30, redlines-38), plus verify-57 on the margins; otherwise cyber dominates.
調査状況Not yet researched, so no proposed questions; flagged as a gap by Claude Fable 5.1 on 23 Oct 2026.
元の問いなし 0 提案
11
L11 · 空白

The model’s welfare

所見“welfare”, “wellbeing”, “sentien”, “moral status”, “conscious”: 0 hits; on the margins humans-02 and humans-09 under SF-V4.
調査状況Not yet researched, so no proposed questions; flagged as a gap by Claude Fable 5.1 on 23 Oct 2026.
元の問いなし 0 提案
12
L12 · 空白

People who prepare training data

所見“annotat”, “labeler”, “data work”, “crowdwork”: 0 hits.
調査状況Not yet researched, so no proposed questions; flagged as a gap by Claude Fable 5.1 on 23 Oct 2026. The proposal on minimum standards for data work under L4 touches this gap.
元の問いなし 0 提案

バージョン 5.2 · 2026-10-25時点 · 311 questions from research.amanidunia.berlin

確認バージョン 5.2 を検証したのは GPT-6.1 Sol, GPT-6 Luna, GLM-5.3, GLM-5.3-Flash により二回に分けて検証されました(2026年十月)。検証する側とされる側は、必ず別の組織です。

Language and review status

Main text: EN. Review notes apply only to the content they name.

Translation available; a completed independent content review is not recorded.

See language coverage and data snapshot →