Habryka comments on Understanding the diffusion of large language models: summary