Year: 2026
Type: article
Abstract: This study investigates the reliability trap in Google Lens, an artificial intelligence-driven augmented reality translation tool. Viewing translation as human-computer interaction, this qualitative field study evaluates artificial intelligence algorithms in the Polish-to-English direction, for so-called incidental texts (text encountered in public spaces, such as signs, notices, and instructions). Analysis indicates that while neural networks deliver high grammatical fluency, their spatial context-blindness to physical geosemiotics causes severe pragmatic pitfalls affecting users, including safety, as well as legal and socio-economic risks. To address this, the paper advocates that traditional equivalence-based quality assessment must give way to a risk management paradigm. Ultimately, operating safely within the augmented linguistic landscape would entail fostering machine translation literacy and risk-aware artificial intelligence interface design.
Author: Michał Organ
Institution: University of Rzeszów
Language: English
Subfield: Language and Linguistics
Field: Arts and Humanities
Domain: Social Sciences
Open Access status: diamond
OpenAlex ID: W7234519154
