Nathan Helm-Burger comments on Capabilities and alignment of LLM cognitive architectures