METHOD OF DESCRIPTION OF ART OBJECTS USING ARTIFICIAL INTELLIGENCE FOR PEOPLE WITH VISUAL IMPAIRMENTS

Authors

DOI:

https://doi.org/10.31891/2219-9365-2025-81-23

Keywords:

artificial intelligence, generative artificial intelligence, description of art objects, prompt, Gemini and ChatGPT language models

Abstract

The proposed method of describing art objects using artificial intelligence for people with visual impairments allows controlling the process of generating the description by artificial intelligence, turning it into a reliable and adaptive communication tool. A clear definition of the task, context, and specific instructions prevents AI “hallucinations” and ensures factual correctness, while a description of the role, communication style, and desired tone of response allows generating an emotionally colored, personalized description that takes into account, for example, the focus of women on emotions and men on structure. The structured response format provides convenience for further automatic conversion of text into multimodal content (Braille or audio recording), reducing cognitive load and making the perception of art intuitive and comfortable.

A comparative analysis of the descriptions of Jan Vermeer's painting "Girl with a Pearl Earring" by Gemini and ChatGPT demonstrates the successful adaptation of both models to the needs of people with visual impairments, but with different stylistic strategies. A common feature is the focus on the contrast between the figure and the dark background, as well as on the pearl earring as a central element. When adapted for women, both models use an emotional-sensory and poetic approach: Gemini emphasizes aesthetic harmony, mystery and direct presence, and ChatGPT - on psychological immediacy, silence and intimacy. In contrast, in the descriptions for men, both models move to an objective, technical analysis: Gemini provides a more conceptual and professional description, focusing on dynamic tension and chiaroscuro techniques, while ChatGPT offers a more balanced, detailed-descriptive analysis of compositional elements and statics. Thus, Gemini exhibits a more pronounced polarization of styles (subjective/artistic vs. technical/conceptual), while ChatGPT uses a more consistent and detailed narrative strategy in both gender adaptations.

Published

2025-02-27

How to Cite

VOICHUR О., & BOYARCHUK А. (2025). METHOD OF DESCRIPTION OF ART OBJECTS USING ARTIFICIAL INTELLIGENCE FOR PEOPLE WITH VISUAL IMPAIRMENTS. MEASURING AND COMPUTING DEVICES IN TECHNOLOGICAL PROCESSES, (1), 195–201. https://doi.org/10.31891/2219-9365-2025-81-23