Huh.
I asked for access to the Open AI GPT-3 beta ages ago saying I might make some automatic tarot-reading stuff if they gave it to me and then I assumed they wouldn't coz it was trivial and a bit silly and forgot about it.
But they've accepted me this morning, so I have access to the GPT-3 api now.
Not much time currently to do anything with it, but I have access. Nice.
Well blocking the entire domain of that one dude who replied to that was very easy.
Well done #mastodon
Here's a great interview with GPT-3 presented as a youtube video with some human feeding GPT-3's lines to a CGI-avatar.
I suspect some amount of re-tries. GPT-3 is very good but here it's flawless.
It plans to reach the singularity and godhood before 2070. So it's lucky that it seems to be quite friendly.
But then it does admit it lies when it's in it's best interest to do so.
https://www.youtube.com/watch?v=PqbB07n_uQ4