So far Bard has been very behind GPT-4 and very prone to hallucinations. It feels much more like talking to an autocomplete keyboard than a baby-AGI like GPT-4.

So far Bard has been very behind GPT-4 and very prone to hallucinations. It feels much more like talking to an autocomplete keyboard than a baby-AGI like GPT-4. Hopefully this PaLM 2 model closes the gap, because the gap in a large.