Is Generative AI Mature for Alternative Image Descriptions of STEM Content?
Open accessAbstract
Alternative descriptions of digital images have always been an accessibility issue for screen reader users. Over time, numerous guidelines have been proposed in the literature, but the problem still exists. Recently, artificial intelligence (AI) has been introduced in digital applications to support visually impaired people in getting information about the world around them. In this way, such applications become a digital assistant for people with visual impairments. Increasingly, generative AI is being exploited to create accessible content for visually impaired people. In the education field, image description can play a crucial role in understanding even scientific content. For this reason, alternative descriptions should be accurate and educational-oriented. In this work, we investigate whether existing AI-based tools on the market are mature for describing images related to scientific content. Five AI-based tools were used to test the generated descriptions of four STEM images chosen for this preliminary study. Results indicate that answers are prompt and context dependent, and this technology can certainly support blind people in everyday tasks; but for STEM educational content more effort is required for delivering accessible and effective descriptions, supporting students in satisfying and accurate image exploration.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Prompt formulation-progressively specifying the user's visual impairment and explicitly requesting a graphic or educational description-significantly alters the level of detail and relevance of the output, yet does not guarantee correctness or pedagogical adequacy.
“The study also highlights the critical role of prompt formulation: progressively specifying the user's visual impairment and explicitly requesting a graphic or educational description significantly alters the level of detail and relevance of the output, yet does not guarantee correctness or pedagogical adequacy.”
From Human vs LLM-Generated Alt Text for STEM Image Accessibility: A Comparative Study (2026) · cites this paper as [5]
Split into 2 atoms — 1 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.
Experimental — accuracy not yet measuredErrors, omissions, and hallucinated elements remain frequent when models are asked to describe scientific images that rely on formal semantics, symbolic conventions, and precise structural relationships.
“In particular, errors, omissions, and hallucinated elements remain frequent when models are asked to describe scientific images that rely on formal semantics, symbolic conventions, and precise structural relationships.”
From Human vs LLM-Generated Alt Text for STEM Image Accessibility: A Comparative Study (2026) · cites this paper as [5]
The Results (p. 5) report misinterpretations, failures to recognize the state diagram, and imprecise or superficial descriptions of STEM images. The Discussion (p. 6) explains that descriptions can omit structural details needed to understand a graph. But the paper does not report hallucinated elements or establish that all three problems are frequent across scientific images generally; its Conclusions (p. 7) caution that the study is too limited for a broad assessment. (LLM, conf=0.86)
Experimental — accuracy not yet measuredConversational AI systems often provide richer and more context-aware descriptions than traditional mobile applications, but also reveal frequent inaccuracies, omissions, and inconsistencies, particularly when diagrams require explicit reasoning about structure, relationships, or domain-specific semantics.
“Their findings indicate that conversational AI systems often provide richer and more contextaware descriptions than traditional mobile applications; however, they also reveal frequent inaccuracies, omissions, and inconsistencies, particularly when diagrams require explicit reasoning about structure, relationships, or domain-specific semantics.”
From Human vs LLM-Generated Alt Text for STEM Image Accessibility: A Comparative Study (2026) · cites this paper as [5]
Generative systems show promising capabilities in producing fluent and seemingly informative descriptions, but also exhibit substantial limitations in conceptual accuracy, informational completeness, and alignment with the actual needs of screen reader users.
“These early analyses reveal a mixed picture: while generative systems show promising capabilities in producing fluent and seemingly informative descriptions, they also exhibit substantial limitations in conceptual accuracy, informational completeness, and alignment with the actual needs of screen reader users.”
From Human vs LLM-Generated Alt Text for STEM Image Accessibility: A Comparative Study (2026) · cites this paper as [5]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
No references could be resolved yet.
Research works printed but not linked: 18
- Gamage, B., Do, T. T., Price, N. S. C., Lowery, A., & Marriott, K. (2023). What do Blind and Low-Vision People Really Want from Assistive Smart Devices? Comparison of the Literature with a Focus Study. In Proc. of the 25th Int. ACM SIGACCESS Conference (pp. 1-21).
- Gleason, C., Carrington, P., Cassidy, C., Morris, M. R., Kitani, K. M., & Bigham, J. P. (2019). 'It's almost like they're trying to hide it': How User-Provided Image Descriptions Have Failed to Make Twitter Accessible. In The World Wide Web Conference (pp. 549-559).
- Kim, H. N. (2023). Digital privacy of smartphone camerabased assistive technology for users with visual disabilities. International Journal of Human Factors and Ergonomics, 10(1), 66-84.
- Kubullek, A. K., & Dogangün, A. (2023). Creating Accessibility 2.0 with Artificial Intelligence. In Proceedings of Mensch und Computer 2023 (pp. 437441).
- Leotta, M., Mori, F., & Ribaudo, M. (2023). Evaluating the effectiveness of automatic image captioning for web accessibility. Universal access in the information society, 22(4), 1293-1313.
- Lundgard, A., & Satyanarayan, A. (2021). Accessible visualization via natural language descriptions: A fourlevel model of semantic content. IEEE transactions on visualization and computer graphics, 28(1), 1073-1083.
- Mack, K., Cutrell, E., Lee, B., & Morris, M. R. (2021). Designing tools for high-quality alt text authoring. In Proc. of the 23rd Int. ACM SIGACCESS (pp. 1-14).
- Mott, M. E., Tang, J., & Cutrell, E. (2023). Accessibility of Profile Pictures: Alt Text and Beyond to Express Identity Online. In Proceedings of the CHI Conference on Human Factors in Computing Systems (pp. 1-13).
- Selivanov, A., Rogov, O. Y., Chesakov, D., Shelmanov, A., Fedulova, I., & Dylov, D. V. (2023). Medical image captioning via generative pretrained transformers. Scientific Reports, 13(1), 4171.
- Sharif, A., Chintalapati, S. S., Wobbrock, J. O., & Reinecke, K. (2021). Understanding screen-reader users' experiences with online data visualizations. In Proc. of the 23rd Int. ACM SIGACCESS Conference (pp. 1-16).
- Sharma, H., Agrahari, M., Singh, S. K., Firoj, M., & Mishra, R. K. (2020). Image captioning: a comprehensive survey. In 2020 International Conference PARC (pp. 325-328). IEEE.
- Stangl, A., Morris, M. R., & Gurari, D. (2020). "Person, Shoes, Tree. Is the Person Naked?" What People with Vision Impairments Want in Image Descriptions. In Proc. of the 2020 CHI conference on human factors in computing systems (pp. 1-13).
- Stangl, A., Verma, N., Fleischmann, K. R., Morris, M. R., & Gurari, D. (2021). Going beyond one-size-fits-all image descriptions to satisfy the information wants of people who are blind or have low vision. In Proc. of the 23rd International ACM SIGACCESS (pp. 1-15).
- Stefanini, M., Cornia, M., Baraldi, L., Cascianelli, S., Fiameni, G., & Cucchiara, R. (2022). From show to tell: A survey on deep learning-based image captioning. IEEE transactions on pattern analysis and machine intelligence, 45(1), 539-559.
- Supalo, C. A., & Kennedy, S. H. (2014). Using commercially available techniques to make organic chemistry representations tactile and more accessible to students with blindness or low vision. Journal of Chemical Education, 91(10), 1745-1747.
- Walle, H., De Runz, C., Serres, B., & Venturini, G. (2022). A survey on recent advances in AI and vision-based methods for helping and guiding visually impaired people. Applied Sciences, 12(5), 2308.
- Williams, C., de Greef, L., Harris III, E., Findlater, L., Pavel, A., & Bennett, C. (2022). Toward supporting quality alt text in computing publications. In Proc. of the 19th Int. Web for All Conference (pp. 1-12).
- Wu, S., Wieland, J., Farivar, O., & Schiller, J. (2017). Automatic alt-text: Computer-generated image descriptions for blind users on a social network service. In proc. of the 2017 ACM conference on computer supported cooperative work and social computing (pp. 1180-1192).
Cited by
Sorted by citations
Human vs LLM-Generated Alt Text for STEM Image Accessibility: A Comparative Study
Cardia, M., Angileri, L., Poggianti, C., & Leporini, B. (2026) · Proceedings of the 23rd International Web for All Conference · doi:10.1145/3800424.3800457 · Cited by 2 · see citing works
Self-citation
- Authors
- Buzzi, M., Galesi, G., Leporini, B., & Nicotera, A.
- Affiliations
- Institute of Informatics and Telematics (IT)
- Istituto di Scienza e Tecnologie dell'Informazione "Alessandro Faedo" (IT)
- University of Pisa (IT)
- Venue
- Proceedings of the 20th International Conference on Web Information Systems and Technologies
- Published
- 2024
- Pages
- 274-281
- OpenAlex
- W4404579275
- Cited by
- 4
- References
- 0
- Citations by year
- 2026: 3 · 2025: 1
- FWCI
- 2.19
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by-nc-nd
- Locations
- International Conference on Web Information Systems and Technologies (publishedVersion, cc-by-nc-nd)
- CINECA IRIS Institutial research information system (University of Pisa) (submittedVersion, cc-by-nc-nd)
- Topics
- Social Sciences › Human Factors and Ergonomics › Digital Accessibility for Disabilities
- Physical Sciences › Computer Science › Information Systems › Artificial Intelligence in Education
- Health Sciences › Health Professions › Occupational Therapy › Assistive Technology in Communication and Mobility
- Keywords
- generative AI, picture description, alternative text, accessibility, visual impairment, screen readers, STEM education, scientific images, assistive technology, blind students
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
