LEAD-ME Seminar "Media Accessibility in the Age of Artificial Intelligence"
Media Accessibility in the Age of Artificial Intelligence
Abstracts
AI and Digital Accessibility - Opportunities and Pitfalls
In this keynote, I start by reviewing how AI has been employed in recent years in the field of accessibility. This will be followed by a reflection on the ways that AI can leverage digital accessibility and the ways that AI hinder digital accessibility.
Empowering Inclusivity: How AI is Revolutionizing Accessibility
In recent years, artificial intelligence (AI) has emerged as a transformative force in enhancing accessibility for people with disabilities. This talk explores the profound impact of AI technologies in creating a more inclusive world. We will delve into the development and implementation of automatic subtitling systems that enable real-time transcription of spoken words, significantly benefiting the hearing-impaired community. Additionally, we'll examine how AI-driven audio descriptions enrich the visual media experience for the visually impaired by providing detailed narrations of on- screen activities. The advancements in image recognition technology will be highlighted, showcasing how AI can identify and describe images, making digital content more accessible. Through real-world examples and case studies, this presentation will illustrate how these innovations not only foster inclusivity but also pave the way for a future where accessibility barriers are systematically dismantled by AI. We will discuss the exciting potential of AI in creating an accessible digital landscape for all.
One size fits one - accessibility through preferences
In 1992, all state-run special schools were closed with the exception of schools for sign language students. The ideology was that special education should take place in a classroom setting together with peers at the local school. In the compulsory primary school, achieving a ‘school for all’, or one that is fully inclusive, is an important policy goal and part of the official aims behind the system of education for all children in Norway.
TV 2 Schools educational webservice – elevkanalen.no – has for more than 10 years developed an API that governs a very successful preference-dashboard. Teacher set preferences on his/her pupils. Eye-gaze navigation, one- button navigation, two-button navigation etc.; symbol-support; background colurs, fontcolour – It is a user-centric design backed up by national activities in Standards Norway Commitee 607 and ongoing work in ISO/IEC JTC1/SC 36/WG 7 - metadata for individualized accessibility.
To achieve true accessibility – AI is obviously the road ahead. Machine learning that automatically detects the users needs is already a component in our one fits one approach, but we need more. This goes for navigation - AI-driven eye-gaze calibration AI-driven tasks like alt-text to pictures, and captions to film for the blind; text-levels etc. We believe our activities already has a proven potential outside our own organisation in Norway and would love to present them in a Lead-Me - context.
The role of AI in translation teaching and its effect on visually impaired students
In the field of education, the impact of information and communication technology enhanced the learning experience and provided easier access to learning materials including the higher education environment. With the advanced technology, the emergence of artificial intelligence and its fast proliferation influenced both academia and the industry in the area of translation. Most universities started to use artificial intelligence (AI) and machine translation (MT) in teaching translation. Concerning the curricula and teaching materials, newly developed translation applications revolutionized the traditional approach and technology-enhanced curricula have been favored in line with the latest developments. The main question is how these technologies can be integrated into the curricula and more importantly how all students can benefit from this latest technology equally. In planning the lessons and forming the curricula, it is essential to take into consideration all types of learners with the ultimate goal of creating an inclusive teaching and learning environment.
The main aim of this qualitative research is to take an in-depth look at the conditions of visually impaired students regarding the availability of AI for them, how much they can benefit from it in the process of learning, and how AI affects them. The data is based on the feedback received from visually impaired students who study in the translation and interpreting department in Türkiye. The results obtained from the study will provide a different perspective on the inclusion of AI in education and its effect on visually impaired students for both their digital and professional skills.
Accelerating Productivity for Media Accessibility Services - An overview of AI for subtitles and audio description use cases
Machine learning methods and artificial intelligence are revolutionising our personal and professional environments, as well as the world of audiovisual translation and media accessibility. Thanks to interfaces with AI applications, the creation of accessible media production and its workflows transformed enormously in recent years. The most prominent examples are speech recognition for automatic transcription in the field of intralingual subtitling, machine translation for interlingual subtitling, and artificial voices for audio description, to name but a few of the advances already in industrial use. However, generative AI in the form of language models and prompt engineering can unleash even more unimagined potential to further increase the already advanced productivity in the near future. These developments can go a long way towards addressing the challenges of mass media accessibility and language transfer. But what does this mean for the quality of future media accessibility services? What ethical questions do these technologies raise? And how do these crucial changes and issues affect the everyday work of audiovisual translators and beyond?
This presentation aims to provide an overview and basis for discussion. It focuses on the implications of the latest developments in AI technology that are already being used or are about to become more relevant in the area of media accessibility.
Inclusive Innovation: Understanding the Experiences of Individuals with Visual Impairments Using Artificial Intelligence Tools
Technological advancements, particularly in artificial intelligence (AI), have revolutionized problem-solving processes, offering efficient solutions to various challenges. Despite inherent challenges, the advantages of technology outweigh its drawbacks. For individuals with disabilities, particularly those with visual impairments, technology plays a pivotal role in mitigating barriers, especially in accessing information. This equality in information access between individuals with and without blindness underscores the profound impact of technology. However, the development of such technological solutions often overlooks the needs of individuals with visual impairments and blindness, leading to limited accessibility of these tools for this demographic. Given the rapid evolution of AI tools and their anticipated influence on our future, it becomes imperative to prioritize inclusivity in their development.
Hence, this research endeavors to delve into the experiences of ten adults with visual impairments who actively utilize AI tools in their daily lives. Through this investigation, we aim to elucidate the specific AI tools employed by individuals with visual impairments, their utilization methods, and the purposes they serve. Additionally, we seek to capture their insights and recommendations for developers of such tools. Ultimately, the session will culminate in actionable suggestions tailored for parents and educators, emphasizing the integration of AI technology into the lives of individuals with visual impairments and blindness. By fostering inclusivity in AI development and utilization, we can pave the way for a more equitable future for all.
Artificial Intelligence Bias in Accessible Smart Tourism
This presentation delves into the pervasive issue of algorithmic bias within Artificial Intelligence (AI) applications, particularly as it pertains to the dynamic field of accessible smart tourism. With the tourism industry increasingly incorporating AI technologies into its operations, there has been a notable enhancement in efficiency and effectiveness. However, this surge in AI adoption has also brought to the forefront the challenge of bias, which poses significant hurdles for both tourists and tourism providers. To dissect this complex problem, the presentation utilizes illustrative scenarios to shed light on various instances of AI bias within the tourism and hospitality sector. Additionally, it introduces a framework designed to assess the global implications of bias within the realms of smart tourism. The presentation underscores the potential ramifications of bias on tourist experiences across different phases: pre-travel, during travel/on-site, and post-travel, while considering viewpoints from both tourists and providers. It emphasizes that many biases originate from the utilization of AI within online platforms, thereby impacting tourists' information retrieval, decision-making processes, and the marketing strategies employed by tourism providers.
Furthermore, it delves into the broader effects of AI on tourist experiences, outlining the social implications, managerial considerations, and technological implications. This presentation represents a pioneering effort in addressing the issue of AI bias within the context of smart tourism. Moreover, it lays out pathways for future research, spanning topics such as technology adoption and consumer behavior, the economic repercussions for tourism providers, and the regulatory landscape.
Gaze-dependent response activation in dialogue agent for cognitive-behavioral therapy
Recent years has seen significant developments within Artificial Intelligence (AI) and Natural Language Processing (NLP), causing increasing interest for their application to other areas, including psychology and medicine. One example is the use of AI technology to complement the work of psychiatrists and therapists. With the aim of providing additional support to psychiatric patients with limited access to medical staff, our team has developed a therapeutic dialogue agent (Terabot). It operates in the Polish language and is enhanced with text- based emotion and intention recognition. It considers the patient's emotions so it can respond in an 'empathetic' way. Experiments have been conducted during which we observed several problems. In particular, when the patient's utterances were too short or too long, the dialogue system reacted inappropriately, interrupting the patient's answer or causing long pauses. This leads to a decrease in the comfort and natural dialogue flow during therapeutic sessions with patients.
To solve these problems, we propose to improve the dialogue system to be multimodal - using eye-tracking data. We consider the fixation parameters at specific areas of interest of Terabot as crucial. We found that the patient's fixations at the end of each utterance can serve as important indicators of the end of speech. This can help to index the end of the utterance explicitly for the dialogue system. As a result, a better timing for Terabot's speech activation can be determined. Thanks to this approach, Terabot's response time can be individually personalized to suit the patients' different response styles.
Diversity for Accessibility: Multimodal Hands-Free Input System for Virtual Community Interaction and Device Control through XR
Extended reality (XR) interactions offer the potential to foster social networks, entertainment, and education and are being adopted by industry for more efficient and safer task completion. As the technology grows in popularity, the need for more inclusive types of interaction for users with motor disabilities becomes imperative. In our project, we dive into the implications and considerations of the use of multimodal input controls for locomotion, pointing and selection actions in XR environments to achieve a hands-free experience. For this, recently adopted AI tools in XR headsets are being used, such as eye-gaze, head-tracking, voice recognition and facial gestures, allowing users with motor disabilities to better access virtual communities through hands-free inputs.
A first study is being carried out where factors such as performance, flow, immersion, engagement and physical effort are measured against the current control systems. This first study allows us to understand the necessary compromises to achieve a similar experience using multimodal hands-free input. Beyond that, the acquired skillset used for virtual environments can also be translated for controlling physical devices, such as wheelchairs or robot movement. Thus, along with our stakeholders, we are designing a pipeline that allows using XR headsets to collect data (e.g. head-tracking) and allows them to move wheelchairs only by using head movements, having at the same time full vision of their environment and enhanced with virtual cues through the headset. We believe the combination of these tools would allow for more inclusive virtual and real-life experiences for users with motor disabilities
Diversity for Accessibility: Multimodal Hands-Free Input System for Virtual Community Interaction and Device Control through XR
This talk will provide an overview about the transformative potential of Extended Reality (XR) technologies, including Virtual Reality (VR) and Augmented Reality (AR), in enhancing accessibility within cultural heritage sites. Through a series of case studies and project examples, a demonstration on how these immersive technologies can provide a more inclusive experience for all visitors regardless of their physical abilities will be presented. The presentation will also discuss practical strategies for adapting these solutions to ensure that that cultural heritage is accessible by everyone.