Can your Private Dat a be Secured in the age of Machine Learning?
Can we trust artificial intelligence to protect our privacy? LLMs like ChatGPT and Claude are trained with both public and private data. How can such a model be designed with privacy by design? These are the questions po sed by computer security professor at Centrum Wiskunde & Informat ica (CWI) Marten van Dijk in his inaugural lecture. Cryptographic met hods like Differential Privacy and PAC Privacy offer partial solution s, but they have limitations. Can these be formally described mathema tically? And can training algorithms be adapted to maintain accuracy while guaranteeing privacy?
Besides privacy, poison and evasion attacks also play a role in undermining the security of AI. There is a need for rigorous comparison of defense mechanisms, something that is often lacking. Fairness by design is also crucial: it has been ma thematically demonstrated that not all fairness definitions are compa tible. How can LLMs still be used fairly and responsibly in decision- making?
Ultimately, we strive for Trustworthy AI: systems that are accurate, secure, fair, and explainable, according to Van Dijk. DESCRIPTION: Can your Private Data be Secured in the age of Machine Le arning? Can we trust artificial intelligence to protect our privacy? LLMs like ChatGPT and Claude are trained with both public and private data. How can such a model be designed with privacy by design? These are the questions posed by computer security professor at Centrum Wi skunde & Informatica (CWI) Marten van Dijk in his inaugural lectu re. Cryptographic methods like Differential Privacy and PAC Privacy o ffer partial solutions, but they have limitations. Can these be forma lly described mathematically? And can training algorithms be adapted to maintain accuracy while guaranteeing privacy?Besides privacy, pois on and evasion attacks also play a role in undermining the security o f AI. There is a need for rigorous comparison of defense mechanisms, something that is often lacking. Fairness by design is also crucial: it has been mathematically demonstrated that not all fairness definit ions are compatible. How can LLMs still be used fairly and responsibl y in decision-making?Ultimately, we strive for Trustworthy AI: system s that are accurate, secure, fair, and explainable, according to Van Dijk. END:VEVENT END:VCALENDAR