Anthropic just proved AI isn't getting better

Anthropic just proved AI isn't getting better argues that advanced language models like Opus 5 remain poor at understanding software architecture and improving code quality, challenging claims that current AI development is progressing toward true general intelligence.