I play with Bard about once a week ago so. It is definitely getting better, I fully agree with that. However, 'better' is maybe parity with GPT-2. Definitely not yet even DaVinci levels of capability.

It's very fast, though, and the pre-gen of multiple replies is nice. (and necessary, at current quality levels)

I'm looking forward to its improvement, and I wish the teams working on it the best of luck. I can only imagine the levels of internal pressure on everyone involved!