For International Data Protection Day, I wrote a blog post for LUMI AI Factory where I interviewed Prof. Antti Honkela from the University of Helsinki’s Department of Computer Science.
Prof. Honkela is an expert in differential privacy and synthetic data generation. We discussed the intersection of data protection and artificial intelligence, including:
Differential Privacy: Real-world adoption challenges for small businesses and public services. As Honkela noted, “implementing differential privacy securely is hard” but tools like OpenDP are improving accessibility.
Synthetic Data: Using artificially generated data for teaching and software development. Works best as a preview tool rather than a substitute for real data in accuracy-critical tasks.
Machine Unlearning: The technical feasibility of removing individuals’ data from trained models remains largely theoretical without special training approaches.
Future Directions: Potential breakthroughs with tabular foundation models that could simplify privacy-friendly AI implementation.
Read the full article at LUMI AI Factory.