Structured Output Collapses Answer Diversity Across 44 Language Models
Что произошло
Источник arXiv cs.CL зафиксировал сигнал: arXiv:2607.18476v1 Announce Type: new
Abstract: When a language model must choose one answer from a large space of equally valid options, a format clause — "Reply with JSON only" — changes which answer it chooses. We re-run the One-Word Census (arXiv:2607.12796): 31 wide-answer-space category prompts asked of 44 models, now with the reply requested in JSON — no schema enforcement, no constrained decoding, only the request. Convergence deepens sharply: on the unconstrained "Pick a word" prompt the modal answer rises from 41% to 64% of the pool and distinct answers fall from 52 to 36; mean answer-choice surprisal drops from 1.80 to 1.58 bits. The tax is progressive: six of 44 models move individually (BH-FDR q=.10), all toward the mode, led by the most distinctive models, while the conformist floor is immobile. It is a sharpener, not a re-indexer — the plain-chat modal answer survives in 28 of 31 categories. Defaults are register-indexed: a within-run re-sample (n=20) finds JSON shifts 53% of a model's stable chat defaults, mostly back to the crowd, and installs defaults absent from chat (Claude Fable 5 answers "cerulean" for colour 0% of the time in chat, 100% in JSON). Full-battery controls reveal a register gradient: compression is significant and specific to the answer-delivery formats models are trained to speak (JSON -0.22 bits, p=.0002; XML -0.19, p=.002), absent for YAML and CSV, and reversed for an arbitrary bracket wrapper (+0.13, p=.009) — weighing the mechanism toward tool-use post-training. Enforcing the schema at the decoder (response_format) compresses no further than the request (-0.03 bits): the collapse lives in the model's response to the register, not the decoder. Structured output is how software consumes language models, and that surface is served by a measurably more homogeneous model than the chat surface on which models are evaluated, compared, and chosen.
Почему это обсуждают
Для аудитории COMRAD404 это повод проверить, касается ли тема моделей, агентов, промптов, инструментов или разработки с ИИ. Социальный источник сам по себе не является доказательством, поэтому выводы нужно держать осторожными.
Что подтверждено
| Punkt | Detail |
|---|---|
| Платформа | arxiv |
| Источник | arXiv cs.CL |
| Проверка | https://www.anthropic.com/news |
Что проверить дальше
Нужно открыть первичный источник, документацию продукта, GitHub, блог лаборатории или публикацию автора и отделить факт релиза от реакции сообщества.
Источник: arXiv cs.CL – https://arxiv.org/abs/2607.18476; проверка: https://www.anthropic.com/news