Pathway CEO Says LLMs Are Fundamentally Broken, Unveils New AI Architecture to Replace Them
Summary
Pathway CEO Zuzanna Stamirowska declares large language models fundamentally broken due to architectural limits and unveils Dragon Hatchling, a new AI system promising native memory, continual learning, and abstract reasoning at lower compute costs.
Key Points
- Pathway CEO Zuzanna Stamirowska argues that today's large language models have fundamental architectural limits, including an inability to truly internalize learned information and constraints caused by reasoning through language.
- Pathway is developing a new architecture called Dragon Hatchling, designed to give AI native memory, continual learning, and more abstract reasoning, with the potential to reduce compute costs and improve interpretability and safety.
- Stamirowska reveals that her engineering team has largely stopped writing code themselves, reflecting how AI coding tools like Codex and Claude Code are already transforming the way AI development teams operate.