This paper presents a comprehensive study on assisting new AI programmers in making responsible choices while programming. The research focused on developing a process model, incorporating design patterns, and utilizing an IDE-based extension to promote responsible Artificial Intelligence (AI) practices. The experiment evaluated the effectiveness of the process model and extension, specifically examining their impact on the ability to make responsible choices in AI programming. The results revealed that the use of the process model and extension significantly enhanced the programmers' understanding of Responsible AI principles and their ability to apply them in code development. These findings support existing literature highlighting the positive influence of process models and patterns on code development capabilities. The research further confirmed the importance of incorporating Responsible AI values, as asking relevant questions related to these values resulted in responsible AI practices. Furthermore, the study contributes to bridging the gap between theoretical knowledge and practical application by incorporating Responsible AI values into the centre stage of the process model. By doing so, the research not only addresses the existing literature gap, but also ensures the practical implementation of Responsible AI principles.
MULTIFILE
While the technical application domain seems to be to most established field for AI applications, the field is at the very beginning to identify and implement responsible and fair AI applications. Technical, non-user facing services indirectly model user behavior as a consequence of which unexpected issues of privacy, fairness and lack of autonomy may emerge. There is a need for design methods that take the potential impact of AI systems into account.
DOCUMENT
Many have suggested that AI-based interventions could enhance learning by personalization, improving teacher effective ness, or by optimizing educational processes. However, they could also have unintended or unexpected side-effects, such as undermining learning by enabling procrastination, or reducing social interaction by individualizing learning processes. Responsible scientific experiments are required to map both the potential benefits and the side-effects. Current procedures used to screen experiments by research ethics committees do not take the specific risks and dilemmas that AI poses into account. Previous studies identified sixteen conditions that can be used to judge whether trials with experimental technology are responsible. These conditions, however, were not yet translated into practical procedures, nor do they distinguish between the different types of AI applications and risk categories. This paper explores how those conditions could be further specified into procedures that could help facilitate and organize responsible experiments with AI, while differentiating for the different types of AI applications based on their level of automation. The four procedures that we propose are (1) A process of gradual testing (2) Risk- and side-effect detection (3) Explainability and severity, and (4) Democratic oversight. These procedures can be used by researchers and ethics committees to enable responsible experiment with AI interventions in educational settings. Implementation and compliance will require collaboration between researchers, industry, policy makers, and educational institutions.
DOCUMENT
In dit project ontwikkelt het HvA-lectoraat Responsible IT in co-creatie met Digital Agency Fonk een werkend prototype van een innovatieve educatieve AI-applicatie, die de taalvaardigheid van kinderen en ouders vergroot. Onderdeel van deze applicatie is een functionaliteit voor taalvereenvoudiging op basis van AI. Dit software-onderdeel analyseert tijdens het lezen het AVI niveau van de lezers en past het verhaal hier automatisch op aan. Met audio- en spraakanalyse worden fouten in o.a. uitspraak, grammatica en woordbegrip gedetecteerd, en het niveau van de tekst automatisch verhoogd of te verlaagd. Door de moeilijkheidsgraad van de tekst langzaam te verhogen wordt de leesvaardigheid verbeterd.
De maatschappelijke discussies over de invloed van AI op ons leven tieren welig. De terugkerende vraag is of AI-toepassingen – en dan vooral recommendersystemen – een dreiging of een redding zijn. De impact van het kiezen van een film voor vanavond, met behulp van Netflix' recommendersysteem, is nog beperkt. De impact van datingsites, navigatiesystemen en sociale media – allemaal systemen die met algoritmes informatie filteren of keuzes aanraden – is al groter. De impact van recommendersystemen in bijvoorbeeld de zorg, bij werving en selectie, fraudedetectie, en beoordelingen van hypotheekaanvragen is enorm, zowel op individueel als op maatschappelijk niveau. Het is daarom urgent dat juist recommendersystemen volgens de waarden van Responsible AI ontworpen worden: veilig, eerlijk, betrouwbaar, inclusief, transparant en controleerbaar.Om op een goede manier Responsible AI te ontwerpen moeten technische, contextuele én interactievraagstukken worden opgelost. Op het technische en maatschappelijke niveau is al veel vooruitgang geboekt, respectievelijk door onderzoek naar algoritmen die waarden als inclusiviteit in hun berekening meenemen, en door de ontwikkeling van wettelijke kaders. Over implementatie op interactieniveau bestaat daarentegen nog weinig concrete kennis. Bekend is dat gebruikers die interactiemogelijkheden hebben om een algoritme bij te sturen of aan te vullen, meer transparantie en betrouwbaarheid ervaren. Echter, slecht ontworpen interactiemogelijkheden, of een mismatch tussen interactie en context kosten juist tijd, veroorzaken mentale overbelasting, frustratie, en een gevoel van incompetentie. Ze verhullen eerder dan dat ze tot transparantie leiden.Het ontbreekt ontwerpers van interfaces (UX/UI designers) aan systematische concrete kennis over deze interactiemogelijkheden, hun toepasbaarheid, en de ethische grenzen. Dat beperkt hun mogelijkheid om op interactieniveau aan Responsible AI bij te dragen. Ze willen daarom graag een pattern library van interactiemogelijkheden, geannoteerd met onderzoek over de werking en inzetbaarheid. Dit bestaat nu niet en met dit project willen we een substantiële bijdrage leveren aan de ontwikkeling ervan.
Denim Democracy from the Alliance for Responsible Denim (ARD) is an interactive exhibition that celebrates the journey and learning of ARD members, educates visitors about sustainable denim and highlights how companies collaborate together to achieve results. Through sight, sound and tactile sensations, the visitor experiences and fully engages sustainable denim production. The exhibition launches in October 2018 in Amsterdam and travels to key venues and locations in the Netherlands until April 2019. As consumers, we love denim but the denim industry, like other sub-sectors in the textile, apparel and footwear industries, faces many complex sustainability challenges and has been criticized for its polluting and hazardous production practices. The Alliance for Responsible Denim project brought leading denim brands, suppliers and stakeholders together to collectively address these issues and take initial steps towards improving the ecological sustainability impact of denim production. Sustainability challenges are considered very complex and economically undesirable for individual companies to address alone. In denim, small and medium sized denim firms face specific challenges, such as lower economies of scale and lower buying power to affect change in practices. There is great benefit in combining denim companies' resources and knowledge so that collective experimentation and learning can lift the sustainability standards of the industry and lead to the development of common standards and benchmarks on a scale that matters. If meaningful, transformative industrial change is to be made, then it calls for collaboration between denim industry stakeholders that goes beyond supplier-buyer relations and includes horizontal value chain collaboration of competing large and small denim brands. However collaboration between organizations, and especially between competitors, is highly complex and prone to failure. The research behind the Alliance for Responsible Denim project asked a central research question: how do competitors effectively collaborate together to create common, industry standards on resource use and benchmarks for improved ecological sustainability? To answer this question, we used a mixed-method, action research approach. The Alliance for Responsible Denim project mobilized and facilitated denim brands to collectively identify ways to reduce the use of water and chemicals in denim production and then aided them to implement these practices individually in their respective firms.
