The law that relates to the collaborative project is a major challenge, and in the context of AI-related projects, it becomes all the more complicated. Josefine Rembsgård, legal coordinator at AI Innovation of Sweden, has looked at how legal challenges are associated with collaborative projects within AI.
In the collaborative projects carried out at the Data Factory at AI Innovation of Sweden, legal requirements often take centre stage. They involve questions relating to ownership of data, legal responsibility, and, not lastly, personal integrity.
"These are questions that generally affect everyone who works on collaborative projects within AI in one way or another, regardless of sector or professional role, so it's important that they're emphasised,
In order to increase knowledge within this area, Josefine Rembsgård, legal coordinator at AI Innovation of Sweden, conducted a preliminary study during the summer and autumn of 2019 mapping the legal challenges that selected participants perceive as being the greatest with regards to the implementation of collaborative projects within AI. The preliminary study also addresses perceived limitations and conditions in order to share data.
"These are questions that generally affect everyone who works on collaborative projects within AI in one way or another, regardless of sector or professional role, so it's important that they're emphasised, " Josefine explains and continues;
"Since it also involves many questions that can be considered as being relatively new, there are challenges in how to handle them"
One of the greatest challenges that is highlighted in the preliminary study report is the legal questions associated with the General Data Protection Regulation, the GDPR.
The questions that are emphasised include how the legislation should be interpreted and the consequences of not having enough support regarding interpretation from relevant authorities. Other challenges that has been identified is that the legislation can be considered as having limiting effects, as well as questions about how data may be collected.
Other questions that have emerged are: Where is it possible to access reliable data? Who is responsible for ensuring that the data satisfies a certain level of quality? And what does anonymisation entail concretely according to the legislation?
The preliminary study also stresses important challenges relating to the need for pedagogy and mediation when several different parties intend to collaborate and understand each other in order to find solutions to legal questions. The need to work with a more cross-disciplinary approach is also stressed so that people with different areas of expertise can understand each other and learn from each other. For example, there is a great deal to gain when legal and technical experts work in close proximity to each other.
"I think pedagogy, in particular, is incredibly important in collaborative projects in order to prevent misunderstandings in general, and especially so that the parties can understand what legal questions they face and how to solve them.
"By also taking a more cross-disciplinary approach to work and sharing knowledge, we can learn important lessons for future projects."
"In order to avoid the perception that the law is an obstacle and something difficult, I think that it should be a fundamental part of collaborative projects from the very beginning in order to structure the project and also include teams that are experts in these different questions. By also taking a more cross-disciplinary approach to work and sharing knowledge, we can learn important lessons for future projects."
In order to investigate the legal conditions, a great deal of work is required at the start of the AI projects that are implemented at the Data Factory. It concerns terms and conditions for sharing of data, how data will be used in the project, the purpose of the use and by whom it will be used. The responsibility for personal integrity is of greatest importance, particularly in relation to the data sets that are sensitive in nature, such as medical data. In addition, the partners' individual conditions must be determined and the legal requirements established in preparation for the collaboration. These requirements must then be included in agreements.
"It's incredibly exciting to be able to participate and follow the different types of collaborative projects, and how the parties greatly improve in collaborating to solve these common challenges. By trying to highlight perceived problems, we're able to more easily address the questions and find answers together, and this preliminary study is an attempt to do just that," Josefine concludes.
The preliminary study report constitutes an objective overall summary of what the participating representatives from partner companies have raised during each interview. The preliminary study report does not reflect the author's own opinions. Participating representatives have had the opportunity to study the material before it was published.
The report was drafted exclusively for partners of AI Innovation of Sweden. If you are a partner and are interested in obtaining the full report, please contact Josefine Rembsgård: firstname.lastname@example.org