If that’s true, Google is doing themselves a huge PR disservice by exposing the Bard they have exposed instead of something closer to the internal one.

Yes, I understand the resources and scaling issue, so (1) Google should recognize that this is literally for all the marbles, and (2) done a slower roll-out with its best foot forward if it needed to rush, while at the same time putting as much money as necessary into assuring it could scale out the good model widely.

Google is a trillion+ dollar company, and is facing an existential threat to its core business (because if someone else’s AI is how people interact with the web, the Google’s ads business evaporates with no replacement; if Google is at least competitive on AI, it may still lose its ads business, but it will have something.) Now is not the time to be cheap.

That’s cool.

When are they planning to sunset it?

If you tell us that then we know they’re actually planning to release it.

:)

Different people asked Bard about it: https://twitter.com/killedbygoogle/status/163831100502438707...

or

"It is currently uncertain when Google Bard will be shutdown, as Google has not announced a specific date. However, given the recent announcement that Google will be shutting down its AI-powered writing tool after less than six months since its launch, it is likely that Google Bard will be shutdown within the next year."

not true. source: I work at Google, and have been using internal bard since they launched it for us internally.

Isn't the internal version of "copilot" (for Cider) pretty good, though?

yeah its pretty useful, especially since its trained on our internal Google3 codebase. I use it multiple times a day on Cider-V.

I have not used copilot from github yet so I dont really have a point of comparison tho.

are you using Big Bard? You have to search for it, it has a lot more parameters.