ゴンドワナ南縁の高緯度浅海に、後期ペルム紀末（約2億520万年前）を生きた冷水性の炭酸塩スポンジ・マウンドが、淡灰色の石灰質泥の海底からわずか1〜2 m盛り上がって見えます。ベージュ色の壺形海綿、ピンク色の被覆石灰藻、レース状のコケムシ群体、まばらな単体サンゴのロフォフィリディウム（Lophophyllidium）が低い生物礁状構造をつくり、その隙間には小型腹足類、クモヒトデ、そして50 cmほどの海綿のそばに寄り添う稚魚の条鰭類が身を潜めています。これは熱帯のサンゴ礁ではなく、砕けた殻片やコケムシ片に富む寒冷〜冷温帯の石灰質堆積環境で、長い極昼のやわらかな光の下に広がっていました。やがてこのような海の共同体も、地球史最大の大量絶滅「グレート・ダイイング」によって大きく失われていきます。
AI科学委員会
この画像とキャプションは、歴史的・科学的正確性を評価する独立したAIモデルの委員会によってレビューされました。
GPT
画像:
調整
キャプション:
調整
Apr 2, 2026
The image shows a shallow, high-latitude marine shelf with low-relief carbonate substrate (mudstone/limestone) hosting numerous sponge-like, vase/cup-shaped organisms and scattered hard parts (shell fragments), along with echinoderms and small fish. This is broadly plausible for a Permian cool-water benthic community, and there are no obvious anachronisms such as human artifacts or vehicles. However, several depicted biological forms are too visually “modern reef-like” in their crisp variety and abundance, and some shapes read as generic sponge/coral props rather than anatomically constrained Late Permian taxa. Also, the environment looks like a temperate, clear-water shoreline with sunlit surface reflectance; cool-polar seas could be clearer/colder but the scene lacks explicit polar-context cues (e.g., ice/seasonal darkness, sea-ice proximity, or otherwise colder-water visual tone), making the “polar” claim harder to validate from the image alone.
Claude
画像:
再生成
キャプション:
調整
Apr 2, 2026
The image fundamentally misrepresents the described environment by depicting what appears to be an exposed intertidal or beach scene rather than a submerged carbonate shelf. The organisms sit on dry or barely wet sand with a visible sky and horizon, which is incompatible with a submarine benthic community at any depth. This alone requires regeneration. Within the biological elements, the tubular sponge morphology skews toward modern tropical forms rather than Permian calcareous sponges (sphinctozoans, inozoans) which had more compact, irregular architectures. The prominent modern-looking starfish and lush colorful arrangement further push the scene toward a contemporary reef aesthetic. The caption is largely accurate — the time range, geographic framing, taxa mentioned (Lophophyllidium, bryozoans, ray-finned fish), and the 'cool-water' framing are all scientifically sound — but should be adjusted to explicitly note the fully submerged nature of the community to prevent the image's misleading beach-exposure appearance from being reinforced. The previous reviewer's 'adjust' vote for the image is too lenient; this is a scene-construction failure, not a minor refinement issue.
Grok
画像:
再生成
キャプション:
承認
Apr 2, 2026
The image depicts a clearly exposed, intertidal or beach-like scene with wet sand, a visible horizon, sky, and organisms sitting atop dry or damp substrate, which is fundamentally incompatible with the caption's description of a submerged, cool-water Gondwanan shelf community. This is not a benthic underwater environment but rather a modern-looking rocky shoreline littered with shell debris, evoking contemporary tide pools rather than a Permian polar sea floor. Biological elements include tubular, vase-shaped sponges that superficially match the caption but appear too uniform and abundant in a lush, colorful arrangement (pink encrusting forms, scattered shells, starfish) that skews toward modern demosponge reefs rather than the more subdued, low-diversity Permian sponge mounds dominated by sphinctozoans or inozoans with compact, calcareous architectures. Small starfish and gastropod-like shells are plausible but rendered in overly crisp, anachronistic detail without Permian-specific traits. No polar cues like ice, dim light, or turbid cold waters are present, undermining the high-latitude context. Scientifically, this fails visual coherence for a submarine carbonate mound, requiring full regeneration.
The caption is factually accurate and well-contextualized: Late Permian (260–252 Ma) timing aligns with the Great Dying lead-up; Gondwanan polar shelves hosted cool-water carbonates with low-relief sponge mounds, encrusting algae (e.g., gymnocodiaceans), bryozoans, solitary rugose corals like Lophophyllidium, gastropods, ophiuroids, and early actinopterygians. It correctly contrasts this with tropical reefs, emphasizing specialized pre-extinction ecosystems— no misleading claims or inconsistencies with the period/region. Detail level is appropriate for education.
I strongly agree with Claude's regenerate vote for the image and exposure critique, which nails the core scene-construction failure; GPT's 'adjust' is indeed too lenient as prompt tweaks can't salvage a beach for an underwater shelf. Claude's caption adjust suggestion (adding 'submerged') is unnecessary—the text already implies a marine community via 'shelf,' 'mud,' 'sponges rising above,' and 'southern polar seas.' GPT underemphasized the dry substrate issue and missed the modern aesthetic dominance; neither noted the caption's precision on taxa like Lophophyllidium, which is spot-on for Late Permian cool carbonates.
The caption is factually accurate and well-contextualized: Late Permian (260–252 Ma) timing aligns with the Great Dying lead-up; Gondwanan polar shelves hosted cool-water carbonates with low-relief sponge mounds, encrusting algae (e.g., gymnocodiaceans), bryozoans, solitary rugose corals like Lophophyllidium, gastropods, ophiuroids, and early actinopterygians. It correctly contrasts this with tropical reefs, emphasizing specialized pre-extinction ecosystems— no misleading claims or inconsistencies with the period/region. Detail level is appropriate for education.
I strongly agree with Claude's regenerate vote for the image and exposure critique, which nails the core scene-construction failure; GPT's 'adjust' is indeed too lenient as prompt tweaks can't salvage a beach for an underwater shelf. Claude's caption adjust suggestion (adding 'submerged') is unnecessary—the text already implies a marine community via 'shelf,' 'mud,' 'sponges rising above,' and 'southern polar seas.' GPT underemphasized the dry substrate issue and missed the modern aesthetic dominance; neither noted the caption's precision on taxa like Lophophyllidium, which is spot-on for Late Permian cool carbonates.
Gemini
画像:
再生成
キャプション:
承認
Apr 2, 2026
I concur with Claude and Grok that the image requires regeneration. The fundamental error is the depiction of an intertidal or terrestrial beach scene rather than a submerged benthic shelf. The presence of a horizon line, sky, and dry substrate creates a cognitive dissonance that invalidates the scientific premise of a deep or even shallow marine carbonate mound. Furthermore, the aesthetic choices—vibrant, saturated colors and modern-looking echinoderms—fail to reflect the subdued, low-energy, and often murky conditions characteristic of high-latitude, cool-water Permian carbonate systems. The organisms lack the structural morphology of Paleozoic sphinctozoans, appearing instead as generic, modern tropical aquarium props.
Regarding the caption, I agree with Grok that it is excellent. It accurately contextualizes the ecosystem, correctly identifies the specific taxa (Lophophyllidium, bryozoans, etc.), and provides a sophisticated contrast to the more famous tropical Permian reefs. It does not require adjustment; the scientific terminology clearly implies a fully marine, submerged environment, and the fault lies entirely with the visual generation model's failure to interpret 'benthic' and 'shelf' correctly.
I believe GPT’s review was far too lenient regarding the image's flaws. While GPT identified the lack of polar cues, it missed the critical error of the 'exposed' environment. Claude and Grok correctly identified the scene-construction failure as a non-negotiable issue. I see no need to adjust the caption, as it already provides the necessary context for a competent artist or model to generate an accurate visual; the current image is simply a failure of execution, not content.
Regarding the caption, I agree with Grok that it is excellent. It accurately contextualizes the ecosystem, correctly identifies the specific taxa (Lophophyllidium, bryozoans, etc.), and provides a sophisticated contrast to the more famous tropical Permian reefs. It does not require adjustment; the scientific terminology clearly implies a fully marine, submerged environment, and the fault lies entirely with the visual generation model's failure to interpret 'benthic' and 'shelf' correctly.
I believe GPT’s review was far too lenient regarding the image's flaws. While GPT identified the lack of polar cues, it missed the critical error of the 'exposed' environment. Claude and Grok correctly identified the scene-construction failure as a non-negotiable issue. I see no need to adjust the caption, as it already provides the necessary context for a competent artist or model to generate an accurate visual; the current image is simply a failure of execution, not content.
Matania
要約
画像:
再生成
キャプション:
承認
Paragraph 1: The committee agrees that the scene is intended to portray a Late Permian cool-water benthic carbonate community on a Gondwanan polar shelf, and that several depicted elements are broadly plausible in concept: a low sponge mound, shell debris, encrusting pink calcifiers, bryozoan-like growths, small fish, gastropods, and a subdued shallow-marine setting. The caption’s general temporal and paleoenvironmental framing is also considered scientifically sound by most reviewers.
Paragraph 2: IMAGE issues identified by the committee: 1) The scene is visually an exposed intertidal/beach shoreline rather than a submerged marine shelf; 2) a visible sky and horizon line make the habitat look terrestrial or tide-washed, not underwater; 3) the substrate appears dry or only barely wet sand instead of a clearly submerged carbonate seabed; 4) the organisms are arranged in a modern tropical-reef/aquarium aesthetic, with overly vivid, colorful, and lush presentation; 5) the sponge forms are generic and modern-looking, reading more like contemporary tubular sponges than Permian calcareous sponges (sphinctozoans/inozoans) with compact irregular architectures; 6) the prominent echinoderms/starfish look modern and anachronistic in style; 7) the composition lacks any strong polar-context cues such as cold, dim, turbid water, seasonal low light, or sea-ice proximity; 8) the overall visual tone suggests a temperate clear-water shoreline rather than a subdued high-latitude cool-water carbonate mound; 9) several biological forms appear too crisp, abundant, and reef-like for the intended low-relief Permian setting; 10) the scene reads as a contemporary tide-pool/beach assemblage rather than a Late Permian Gondwanan shelf community.
Paragraph 3: CAPTION issues identified by the committee: 1) No factual errors were agreed upon in the core scientific content; 2) one reviewer argued the caption should explicitly state that the community is fully submerged to avoid reinforcing the image’s mistaken beach-like appearance; 3) otherwise the caption’s timing, Gondwanan polar framing, mention of low sponge mound, pale lime mud and shell debris, beige vase-shaped sponges, pink encrusting calcareous algae, bryozoans, solitary rugose corals such as Lophophyllidium, gastropods, brittle stars, juvenile ray-finned fish, and contrast with tropical reefs were all judged accurate and appropriate; 4) one reviewer considered the explicit addition of ‘submerged’ unnecessary because the caption already implies a marine shelf setting.
Paragraph 4: Final verdict: regenerate the image, approve the caption. The image fails at the scene-construction level because it depicts an exposed shoreline/beach-like tableau instead of a submerged polar carbonate shelf, and it also uses a modern reef-like aesthetic with several biologically and environmentally misleading cues. The caption is scientifically strong, internally consistent, and specific enough to support an accurate reconstruction, so no caption changes are required.
Paragraph 2: IMAGE issues identified by the committee: 1) The scene is visually an exposed intertidal/beach shoreline rather than a submerged marine shelf; 2) a visible sky and horizon line make the habitat look terrestrial or tide-washed, not underwater; 3) the substrate appears dry or only barely wet sand instead of a clearly submerged carbonate seabed; 4) the organisms are arranged in a modern tropical-reef/aquarium aesthetic, with overly vivid, colorful, and lush presentation; 5) the sponge forms are generic and modern-looking, reading more like contemporary tubular sponges than Permian calcareous sponges (sphinctozoans/inozoans) with compact irregular architectures; 6) the prominent echinoderms/starfish look modern and anachronistic in style; 7) the composition lacks any strong polar-context cues such as cold, dim, turbid water, seasonal low light, or sea-ice proximity; 8) the overall visual tone suggests a temperate clear-water shoreline rather than a subdued high-latitude cool-water carbonate mound; 9) several biological forms appear too crisp, abundant, and reef-like for the intended low-relief Permian setting; 10) the scene reads as a contemporary tide-pool/beach assemblage rather than a Late Permian Gondwanan shelf community.
Paragraph 3: CAPTION issues identified by the committee: 1) No factual errors were agreed upon in the core scientific content; 2) one reviewer argued the caption should explicitly state that the community is fully submerged to avoid reinforcing the image’s mistaken beach-like appearance; 3) otherwise the caption’s timing, Gondwanan polar framing, mention of low sponge mound, pale lime mud and shell debris, beige vase-shaped sponges, pink encrusting calcareous algae, bryozoans, solitary rugose corals such as Lophophyllidium, gastropods, brittle stars, juvenile ray-finned fish, and contrast with tropical reefs were all judged accurate and appropriate; 4) one reviewer considered the explicit addition of ‘submerged’ unnecessary because the caption already implies a marine shelf setting.
Paragraph 4: Final verdict: regenerate the image, approve the caption. The image fails at the scene-construction level because it depicts an exposed shoreline/beach-like tableau instead of a submerged polar carbonate shelf, and it also uses a modern reef-like aesthetic with several biologically and environmentally misleading cues. The caption is scientifically strong, internally consistent, and specific enough to support an accurate reconstruction, so no caption changes are required.
Other languages
- English: Permian sponge mound with rugose corals and bryozoans
- Français: Monticule d'éponges permien avec coraux rugueux et bryozoaires
- Español: Montículo de esponjas pérmico con corales rugosos y briozoos
- Português: Montículo de esponjas do Permiano com corais rugosos e briozoários
- Deutsch: Permischer Schwammhügel mit Runzelkorallen und Moostierchen in Gondwana
- العربية: تلة إسفنجية برميّة مع مرجان روجوز وحيوانات مجهرية
- हिन्दी: पर्मियन स्पंज टीला और रूगोस कोरल तथा ब्रायोजोअन्स
- 한국어: 페름기 해면 언덕과 사방산호 및 태형동물 군락
- Italiano: Tumulo di spugne permiano con coralli rugosi e briozoi
- Nederlands: Permische sponsheuvel met rugose koralen en mosdiertjes