I’m specifically looking for insight into how transformers/ embeddings are evolving classical planning techniques.<p>If we make headway here then we are likely to have autonomous agents that are able to determine an outcome of their conversation vs the markovian way they act today!