We are excited to explore our first Co-Creation Workshops for the THEMIS 5.0 project in this second post of our blog series. Today, we will recapitulate the mission of THEMIS 5.0 and how the Co-Creation workshop has provided significant groundwork in aiding that mission. In the first of this blog series, we outlined the methodology of the Co-Creation project, in this post we will overview the entire project before going into the findings in the following series.
THEMIS 5.0
The THEMIS 5.0 project exists to help and guide creators of artificial intelligence (AI), so that they may be better equipped to develop a more trustworthy and human-centric approach in their programming. It is undeniable that AI will become a fundamental feature of our modern life: having developed exponentially in potential and transitioning from merely the realm of imagination to becoming a common element in many households and workplace in only a few years.
Like the internet, it appears probable that even the most ardent technophobe will fail to shun AI, but that does not mean that we can have no say in its development. At the moment, this socially transformational technology is being guided solely by the preferences of Big Tech, and the European Commission has been working towards reorienting the priorities of AI towards the needs of citizens.
The EU has already begun its work in guiding the development of AI, with the Artificial Intelligence Act coming into force on the 1st August this year: with a stated mission to “promote the uptake of human-centric and trustworthy artificial intelligence.” However, this goal cannot be implemented without an adequate understanding of the nature of trustworthiness and human-centric approaches, especially with relation to AI.
This is where THEMIS 5.0 is critical, with a mission “to create an ecosystem through which AI-driven hybrid decision making is in accordance with particular human user decision support needs and moral values.” To fulfil this mission, the THEMIS 5.0 project has thoroughly investigated what trustworthiness and human centric approach means in a real (and not merely academic sense) to relevant stakeholders. Identifying the hopes and concerns of citizens so that they may be factored into the development of AI will be critical in overcoming hostility towards the uptake of this new (and fundamentally exciting) technology and accelerating productivity and innovation in Europe.
Co-Creation 3.1
With that in mind, we can truly understand the role and value of our first Co-Creation activities within the project. Our Co-Creation work seeks to gather research in multiple workshops from a diverse range of stakeholders including AI developers, domain experts, end users and other relevant parties living and working throughout Europe, so that their insights can inform the advancement of the THEMIS 5.0 ecosystem.
Co-creation workshops invite members to actively participate in the creation of findings through shared experiences and contributions. It is uniquely advantageous because it increases the levels of acceptance, trust and insight through participatory and open discussion as well as facilitating ongoing research and adaptation to changing contexts. Additionally, the findings from Deliverable 3.1 will provide the foundation for further research and co-creation projects such as Deliverables 2.1, 2.2 and 4.1.
The Co-Creation workshops focused their investigations on the role of artificial intelligence in the areas of healthcare, port management and disinformation in the media. With healthcare workshops in Denmark and Bulgaria, port management workshops in Spain, the Netherlands and Poland, and media disinformation workshops in Austria, Greece and Lithuania. These eight workshops engaged 188 participants who were likely to interact with AI. Each workshop focused on two points: first, gathering insights of participants' attitudes, concerns and hopes towards AI being implemented in their work and second, exploring their sector’s understanding of the AI trustworthiness variables of fairness, accuracy and robustness.
Findings
These workshops produced a plethora of results, which we shall explore in the following weeks. Here is a preview of the kind of information that was collected. The first session dealt with three scenarios, these being:
1) General attitudes towards the implementation of AI tools in their work
2) What flaw in the AI tool the participants were least likely to accept
3) How the participants would handle a situation where their professional judgement was contrary to the AI tools suggestion.
From these scenarios and the following discussions, the first session of the Co-Creation project identified four topic of discussion:
Responsibility
Transparency and Accuracy
Cost vs. Efficiency (in healthcare) & Contextual Nuances (in port management and disinformation in media)
Attitudes towards AI
In the second session, when exploring the parameters of trustworthiness, each of the workshops discovered 11 indicators for fairness, 10 for accuracy and 12 for robustness.
We see that there is a great impetus driving the need for the THEMIS 5.0 project and a great treasure of knowledge in the research of our first Co-Creation project, which we look forward to sharing widely in the coming blog posts as we delve deeply into the respective findings in the sectors of healthcare, port management and information in media.
Commentaires