IMVC Webinar Series - The Compression Paradox: Why AI and Humans See the World Differently
- masha675
- Sep 16
- 1 min read
Updated: Oct 27
The first webinar in the IMVC Webinar Series took place with Dr. Ravid Shwartz-Ziv, Assistant Professor and Faculty Fellow at New York University (NYU), delivering a fascinating session on The Compression Paradox: Why AI and Humans See the World Differently.
In this engaging talk, Dr. Shwartz-Ziv explored how foundation models achieve superhuman performance yet perceive the world differently from humans. Using information theory, he revealed a fundamental paradox: AI optimizes for statistical compression, while humans maintain “inefficient” representations that preserve meaning.
Key insights from the webinar include:
How AI excels at broad pattern recognition but struggles with fine-grained conceptual understanding
Examples from multi-agent debates, concept formation, and chess showing the limits of AI reasoning
Why current AI models optimize for efficiency over semantic comprehension and what it means for human-aligned AI
Dr. Shwartz-Ziv is an Assistant Professor and Faculty Fellow at NYU’s Center for Data Science, where he leads cutting-edge research in artificial intelligence, with a particular focus on Large Language Models (LLMs) and their applications. Dr. Shwartz-Ziv’s work spans theoretical foundations and practical implementations, combining academic rigor with industry impact.
Watch the full recording here
Stay tuned for more webinars in the series leading up to the 15th Israel Machine Vision Conference (IMVC) 2026 on April 27, 2026, at Pavilion 10, EXPO Tel Aviv.
