PulseAugur
EN
LIVE 03:21:03

Readers still prefer human translations over AI-generated literary texts

By PulseAugur Editorial · [2 sources] ·

A new study published on arXiv reveals that while AI-generated translations of literary texts are considered "fine" by readers, human translations are still preferred for their immersive quality and clarity. The research involved 15 avid readers comparing human and AI translations of novels in French, Polish, and Japanese into English. Although readers could not reliably distinguish between the two, they favored human translations, noting that AI translation quality varied more within a single book. The study also found that current automatic metrics, including LLM-as-a-judge approaches, fail to accurately reflect reader preferences. AI

IMPACT Suggests human oversight and nuanced evaluation are still critical for high-quality literary translation.

RANK_REASON Academic paper on AI translation quality and reader preference.

Read on arXiv cs.CL →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

Readers still prefer human translations over AI-generated literary texts

COVERAGE [2]

  1. arXiv cs.CL TIER_1 English(EN) · Yves Ferstler, Adam Podoxin, Ty Brassington, Roman Grundkiewicz, Maite Taboada, Marzena Karpinska ·

    AI translation of literary texts is "fine", but readers still prefer human translations

    arXiv:2606.26040v1 Announce Type: new Abstract: AI translation of literary works is increasingly common. While the content may be rendered adequately, we do not know enough about how readers experience it in terms of immersiveness and literary effect, aspects poorly captured by a…

  2. arXiv cs.CL TIER_1 English(EN) · Marzena Karpinska ·

    AI translation of literary texts is "fine", but readers still prefer human translations

    AI translation of literary works is increasingly common. While the content may be rendered adequately, we do not know enough about how readers experience it in terms of immersiveness and literary effect, aspects poorly captured by automatic machine translation metrics or human ev…

RELATED ENTITIES

RELATED TOPICS