Databricks' Yuchen Jin proposes an analogy likening human attention and meetings to the computational cost of burning LLM tokens · Digg