AI based solutions and services appear in proposals across Horizon Europe, from the highly technical approach of the digital, industry and space cluster to the application areas addressing global challenges, as health, environment and transport and the deployment areas of Digital Europe. 

While just about everyone agrees properly addressing AI robustness is essential to a responsible European approach to artificial intelligence, there is little agreement on the best and most consistent way to evaluate it. Evidence of AI Robustness is expected in a 10 page short proposal and vast technical proposals. In multidisciplinary areas, experts may not be prepared to assess AI Robustness. It is clear, that a consistent approach and related guidance is needed. 

This workshop will start the discussion on the issue with a community most likely to face this challenge, but also having the most expertise to address it. The result will be initiatives to share recommendations and good practices to a broader audience.

Agenda

• Welcome and introduction, Tour de table 
• Aspects of AI Robustness and their relevance vis-à-vis Evaluation criteria
• Challenges of evaluating AI Robustness 
• Dos and Don'ts for applicants and evaluators
• Questions and answers