In recent years, text-to-image generative models have advanced rapidly, enabling users to create high-quality, detailed images without the need for traditional artistic skills. These technologies hold great promise for democratizing artistic expression, providing new means of creative exploration and self-empowerment. For individuals with motor impairments or other disabilities, generative AI offers an unprecedented opportunity for self-expression, allowing them to communicate visually in ways that were previously inaccessible. However, the predominant reliance on text input in current image generation platforms poses an important usability barrier. For individuals with cognitive disabilities, as well as for those with limited written articulation skills, the complexity and cognitive demands of crafting effective text prompts can be prohibitive. This usability gap underscores the need for a more inclusive approach to text-to-image AI, one that accommodates diverse cognitive and linguistic abilities while supporting an engaging and empowering user experience. This thesis uses an inclusive design approach to address these accessibility challenges and to reimagine text-to-image generative platforms. Through a detailed analysis of the unique needs of users with cognitive disabilities, combined with insights from iterative user testing, the study presents a framework for developing accessible, adaptive tools in AI-driven creative applications. The resulting solution is a web-based platform that adheres to Web Content Accessibility Guidelines (WCAG) and content simplification principles, facilitating ease of use and maximizing accessibility. The platform offers different approaches to prompt generation, allowing users to choose between structured guidance and open-ended options based on their preferences and needs. This research contributes to the broader conversation on inclusive design in generative AI by demonstrating how an accessibility-focused approach can transform digital tools into platforms for universal creativity that can foster opportunities for self-expression and digital engagement across a diverse range of users.
Negli ultimi anni, i modelli generativi text-to-image hanno fatto grandi progressi, consentendo la creazione di immagini dettagliate e di alta qualità senza la necessità di competenze artistiche. Queste tecnologie offrono l’opportunità di democratizzare l'espressione artistica, fornendo nuovi mezzi per l'esplorazione creativa e l’affermazione personale. Per le persone con disabilità motorie o altre limitazioni, l'IA generativa offre inedite opportunità di espressione, creando nuovi canali comunicativi prima inaccessibili. Tuttavia, la dipendenza da input testuale nelle attuali piattaforme di generazione di immagini rappresenta un importante ostacolo. Per le persone con disabilità cognitive o con limitate capacità di articolazione scritta, la complessità e impegno cognitivo richiesto per creare prompt testuali efficaci possono risultare proibitive. Ciò evidenzia la necessità di un approccio più inclusivo all'IA text-to-image, che tenga conto delle diverse abilità cognitive e linguistiche, garantendo un'esperienza accessibile a tutti. La tesi utilizza un approccio di progettazione inclusiva per affrontare lacune di accessibilità e per ripensare le piattaforme generative text-to-image. Attraverso un'analisi dettagliata delle esigenze degli utenti con disabilità cognitive e i risultati di test iterativi, la tesi propone un quadro per lo sviluppo di strumenti accessibili ed efficaci per le applicazioni creative che utilizzano intelligenza artificiale. La soluzione proposta è una piattaforma web conforme alle linee guida per l'accessibilità dei contenuti web (WCAG) e ai principi di semplificazione dei contenuti, facilitando l’intera esperienza. La piattaforma offre diversi approcci alla generazione di prompt, consentendo agli utenti di scegliere tra una guida strutturata e opzioni aperte in base alle loro preferenze ed esigenze. Questa ricerca contribuisce ad una più ampia conversazione sull’ AI generativa, dimostrando come un approccio inclusivo possa trasformare gli strumenti digitali in piattaforme utili per la creatività di tutti, che favoriscono le opportunità di espressione personale e di coinvolgimento nell’uso di strumenti digitali per una gamma vasta di utenti.
Empowering creativity through text-to-image AI: an inclusive design approach for cognitive disabilities
Predieri, Francesca
2023/2024
Abstract
In recent years, text-to-image generative models have advanced rapidly, enabling users to create high-quality, detailed images without the need for traditional artistic skills. These technologies hold great promise for democratizing artistic expression, providing new means of creative exploration and self-empowerment. For individuals with motor impairments or other disabilities, generative AI offers an unprecedented opportunity for self-expression, allowing them to communicate visually in ways that were previously inaccessible. However, the predominant reliance on text input in current image generation platforms poses an important usability barrier. For individuals with cognitive disabilities, as well as for those with limited written articulation skills, the complexity and cognitive demands of crafting effective text prompts can be prohibitive. This usability gap underscores the need for a more inclusive approach to text-to-image AI, one that accommodates diverse cognitive and linguistic abilities while supporting an engaging and empowering user experience. This thesis uses an inclusive design approach to address these accessibility challenges and to reimagine text-to-image generative platforms. Through a detailed analysis of the unique needs of users with cognitive disabilities, combined with insights from iterative user testing, the study presents a framework for developing accessible, adaptive tools in AI-driven creative applications. The resulting solution is a web-based platform that adheres to Web Content Accessibility Guidelines (WCAG) and content simplification principles, facilitating ease of use and maximizing accessibility. The platform offers different approaches to prompt generation, allowing users to choose between structured guidance and open-ended options based on their preferences and needs. This research contributes to the broader conversation on inclusive design in generative AI by demonstrating how an accessibility-focused approach can transform digital tools into platforms for universal creativity that can foster opportunities for self-expression and digital engagement across a diverse range of users.File | Dimensione | Formato | |
---|---|---|---|
2024_12_Predieri.pdf
solo utenti autorizzati a partire dal 20/11/2025
Dimensione
18.49 MB
Formato
Adobe PDF
|
18.49 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/231407