YES

The transforms these LLMs are built on are not as efficient as they are novel. Without repeatability there won’t be any kind of self learning so a next step in AI won’t work.