BY Paschal Ochang and Irma Poder
The University of Nottingham hosted a co-creational workshop on the 29th -31st January at the Jubilee Hotel & Conferences facilities, bringing together participants of the RAISE project’s case studies located across the UK and Africa. Participants were representatives of small and medium-sized enterprises (SMEs) employing or developing services using generative AI in a diverse range of fields, such as health, education, finance, legal and the creative sectors. The workshop allowed participants to better understand the challenges around integrating measures for responsible research and innovation (RRI) into the use of generative AI.
As a basis for the discussion University of Nottingham professor Bernd Stahl introduced the history and principles of RRI. The idea of RRI is in part to guide research and innovation projects in a way that makes them more open, inclusive and transparent as well as more adaptive to change. The researchers and innovators are asked to consider the impact of the project or product in a forward-thinking capacity. Researchers in the field of RRI have developed frameworks such as the AREA framework to help guide discussions by establishing road signs of potential impacts to look out for. In the EU these include for example ethics, gender equality, governance, open access, public engagement and science education.
To put the idea of RRI into practice, participants were divided into groups of two or three and asked to come up with a pitch for a business idea that uses generative AI. These ideas ranged from sector to sector with pitches for tools for solicitors, alternative medicine, teachers and many more. Once the ideas were made more concrete, participants were then asked to consider potential impacts using the RRI Prompts and Practice Cards. These were practical questions to help participants consider the different ways that the business model could impact society, the environment and more.
These cards are a fun and engaging way to consider the impact that a business may have – even if some of the questions were found to not apply to the particular use case e.g. around potential conflicts or diversity of the team, taking the time to reflect on the many ways that people outside the business may be impacted allowed participants to create more solid pitches.
Other tools exist to consider RRI specifically in the context of AI, and the participants reviewed some of these tools next: the UK Government’s AI Management Essentials, the Assessment List for Trustworthy Artificial Intelligence (ALTAI), and Microsoft’s Responsible AI Impact Assessment Tool. These tools employed a checklist type approach to once again help decision-makers in businesses consider the various impacts that the use of AI may have. Participants were asked to share their feedback on the useability of these checklists and the findings were quite similar across the three different tools. Participants suggested that the tools were useful for facilitating a discussion. However, the formatting of them made some questions difficult to understand or interpret. It was not always obvious which questions need to already be answered at the business brainstorming stage and which aspects of the checklists need to be addressed later on in the development process. Therefore, the possibility to save and come back to answers at a later stage was valued.
Finally, the experience of the day and understanding the practical tools created for supporting RRI implementation were used to provide feedback to the RAISE guidelines. Feedback on the RAISE guidelines highlighted the need for an adaptable, working document that supports SMEs in tracking progress, prioritising relevant aspects, and integrating responsible AI into their development processes. Participants recommended a maturity scoring system, to help assess which aspects of RRI should be prioritised at which stage of development as well as what to prioritise when there is a conflict between the different elements. Based on the different tools used throughout the day, the participants appreciated the more interactive formats and the opportunities to return and update the checklists as the business progresses.
The workshop highlighted the challenges SMEs face in adopting responsible AI practices, including resource constraints around expert knowledge available in a startup that is necessary to answer some of the checklist questions and navigate the existing governance tools, as well as the need for more actionable, SME-focused frameworks. The feedback from the workshop will be used to update the RAISE guidelines and contribute to the ongoing development of responsible AI strategies for SMEs.
