The pilot action in BETTI defines and tests application of an assessment tool for trustworthy AI in smart city/region sectors, supported by a multi-disciplinary/sector AI ethics committee. It is a new initiative, jointly designed by the project, though it also transfers partners’ knowledge.
Overall, public authorities in BETTI have identified the lack of a practical tool that that civil servants can apply as a standard assessment on initiatives for AI deployment within their spheres of influence, in relation to elements of ethical AI.
The pilot action proposal begin from the EU’s High Level Group on AI “Assessment List for Trustworthy AI - ALTAI”. This tool turned guidelines into a practical checklist and on-line tool that helps organisations to self-assess trustworthiness of AI systems under development. ALTAI should be completed with a multidisciplinary team of people with competences or expertise on each of the Trustworthy AI requirements.
The pilot action structure is as follows, with coordination from the Lead Partner and a working group with all participating BETTI partners:
• Partners analyse ALTAI to ascertain appropriateness for their specific needs, for example the extent to which questions of gender can be assessed.
• Partners with existing assessment systems bring in their expertise (PZH from national level tool), as extra elements and tools that could be support ALTAI application (including tools available on the market) are mapped and analysed.
• Each partner set up an ethics committee and adapts / procures matching tools on the specific AI service / tool / initiative selected (see details for the Regional Digital Agenda above).
• In each territory, ALTAI, with integrated measures and the matching tools, is trialled with support from the AI Ethics Committee.
• An interregional evaluation provides recommendations to mainstream application of the assessment tool into policy instruments.