I'm confused. Is this a paper or a blog post? It's basically a play by play of "something we asked an LLM to do". Is this what constitutes research at DeepMind these days?