Summary:
-
Belief in Scaling Laws: Microsoft CTO Kevin Scott asserts that large language model (LLM) “scaling laws” will continue driving AI progress. He emphasizes that we’re not at diminishing returns on scaling up and highlights the exponential growth in AI capabilities as models and computational power increase.
-
Significant Deal with OpenAI: Scott played a key role in the $13 billion technology-sharing deal between Microsoft and OpenAI. He explains that scaling models and training data can lead to substantial improvements in AI, despite some skepticism in the field.
-
Ongoing Debate: The concept of scaling laws remains a cornerstone of OpenAI’s development philosophy, though other researchers have challenged its persistence over time. Scott’s comments reinforce Microsoft’s commitment to advancing AI through increased model size and training data.
!