AI in medicine - EU decision

Gian Battista Parigi

Pavia University Conference �on Artificial Intelligence

Opening speech delivered by the Director of Artificial Intelligence Development Laboratory at Harvard University

19-4-2024

• A section handles limited risk AI systems, subject to lighter transparency obligations: developers and deployers must ensure that end-users are aware that they are interacting with AI (chatbots and deepfakes).
• Minimal risk is unregulated (including the majority of AI applications currently available on the EU single market, such as AI enabled video games and spam filters – at least in 2021; this is changing with generative AI).

​

Prohibited AI systems

• deploying subliminal, manipulative, or deceptive techniques 
• exploiting vulnerabilities 
• biometric categorisation systems 
• social scoring
• assessing the risk of an individual committing criminal offenses 
• compiling facial recognition databases 
• inferring emotions in workplaces or educational institutions
• real-time remote biometric identification (RBI) in publicly accessible spaces for law enforcement,

​

Evolution of clinical research

vs

​

• Unlimited data availability

​

• Data elaboration through predefined algorithms

​

• Results validation through statistical analysis

• Limited data availability (15 TB)
• Data elaboration through unpredictable mental associations
• Results validation through personal experience

Data availability

• the global datasphere will grow to 175 zettabytes by 2025 (1 ZB equals 1 trillion GB)

Hard Disk 5 Mbytes,

IBM 1956

ChatGPT parameters database

Imaginative data elaboration

Evolutionary algorithm

Deep learning

On / Off

0 / 1

Yes, No, Perhaps…