Google Search could be the world's source of truthful information, actively de-prioritizing the dead wood of the internet as they were in the 2000's. Google could still operate a vital layer in the AI future, where humans and bots go to find accurate answers.

LLMs are useful for many creative tasks, but mingling search with dubious LLM proclamations doesn't get me there quicker. For me at least it just introduces a layer of doubt, and more time spent verifying the results.

Bard is better than GPT3.5 (ChatGPT free) and, IMHO, worse than GPT4 (ChatGPT paid).

I find the ChatGPT's interface much better. Bard is really barebones and heavy. Now with GPT store and whatnot, OpenAI is not making it easy for Google at all.

Bard UI feels wrong form me, the same way other search engines feel wrong. I can't put my finger, maybe its because they don't seem to stream the output letter by letter or maybe the UI styling doesn't look like a chat box when it behaves like a chatbox?

OpenAI might have defined how an AI should look and behave.

For me, it was about 'what happens next with the output'. I found it really annoying trying to share or export. Bard behaved as it should. It's a few small things that make the difference.

I find that they both might out pretty similar stuff. In which case, I find myself acting like water ... finding the easiest end point.

We need a meta AI that will query all the other AIs and improve their prompts and summarize the good results

If that occurs, perhaps they can name it “Yo Dawg.”