DALLAS, Oct. 26, 2026 (GLOBE NEWSWIRE) -- Barmeter Technologies LLC has issued a statement highlighting published research involving software engineer Harsha Vardhan Reddy Nooka on the accessibility and functional reliability of front-end code generated by large language models.
The study, “AXIS-Bench: A Pilot Benchmark For Interactivity and Accessibility Conformance in Front-End Code Generation,” was published August 1, 2026, in the International Journal of Artificial Intelligence and Machine Learning. It introduces a pilot benchmark for evaluating whether generated interface components function correctly for keyboard users and assistive technologies, beyond simply resembling a visual design.
The benchmark contains 12 interactive component specifications, including dialogs, tabbed navigation, sortable tables, forms and keyboard-operable drag-and-drop interactions. Each is evaluated against relevant WCAG 2.2 and WAI-ARIA interface expectations using automated accessibility checks and browser-based interaction tests.
In the reported pilot evaluation, a non-semantic baseline passed none of the 12 interactivity and ARIA-conformance tasks. A single-pass large language model condition passed six, while a condition supplied with relevant accessibility-pattern guidance passed all 12. The results are limited to the study’s pilot tasks and should not be interpreted as proof that guideline-grounded AI-generated code will always satisfy accessibility requirements in production.
Barmeter Technologies noted that the findings illustrate the importance of evaluating generated software against practical functional and accessibility requirements, rather than relying on appearance alone. The company’s statement highlights the broader relevance of measurable quality checks as AI-assisted software development becomes more common.
Nooka is a senior front-end engineering professional at Amazon whose experience includes AWS Management Console platform engineering, secure cross-origin communication, performance improvements and integrations involving Amazon Q. His professional background provides context for his interest in dependable web interfaces and front-end engineering. Amazon is identified solely as part of his professional background and is not presented as a sponsor or participant in the research.
The publication presents a pilot benchmark and reports experimental findings; this announcement does not claim that Barmeter Technologies, Amazon or any other organization has commercially deployed AXIS-Bench.
About Barmeter Technologies
Barmeter Technologies LLC is the company issuing this statement on publicly available research concerning software engineering and emerging technology. The statement does not announce a partnership, product launch or research sponsorship.
Research Reference
“AXIS-Bench: A Pilot Benchmark For Interactivity and Accessibility Conformance in Front-End Code Generation.” International Journal of Artificial Intelligence and Machine Learning, Vol. 6, No. 8s (2026), pp. 1390–1402. Published August 1, 2026. DOI: https://doi.org/10.51483/IJAIML.3015
Media Contact
Barmeter Technologies LLC | services@barmetertech.com
A photo accompanying this announcement is available at https://www.globenewswire.com/NewsRoom/AttachmentNg/5a3be84b-34e2-4f97-8235-6f06d18dcfd4
