I strongly believe that by scaling up LLMs we will get there. I write about this here in more detail: https://adam.new/blog/bitter-lesson-ai-cad