Rethinking Knowledge Extraction
Sepp emphasizes the need for new approaches to extract knowledge from existing systems, suggesting that while large language models can combine known concepts, they struggle to generate entirely new ideas. Tim challenges this by exploring the potential of tweaking current methods, hinting at the complexity of program synthesis and the limitations of existing models in inventing novel concepts. The conversation highlights a critical juncture in AI development, questioning whether innovation can emerge from mere combination or if it requires fundamentally new frameworks.In this clip
From this podcast
Machine Learning Street Talk (MLST)
Sepp Hochreiter - LSTM: The Comeback Story?
Related Questions