[aiono]

So it means that throwing data and computing into LLMs won't make them intelligent as opposed many people claiming that will happen. Also they don't play well with the situations that are not in the dataset since they are just extrapolating from what they are trained without any real understanding.