@bence @NathanpmYoung how close does this come to resolving?

If it's sonnet 3.5v2 how does it resolve?

It's possible that this will get resolved based off a technicality - i.e. a video does get posted but without proof of it being executed by Claude. Otherwise a pretty strong No - the first rule of Twitter is that any viral tweet without irrefutable proof in the thread is at least a strong exaggeration.

Is this a new version of sonnet 3.5? Otherwise I'm confused - couldn't anybody reproduce this?

Arb? https://manifold.markets/AdamK/will-an-ai-minecraft-agent-defeat-t-a3b3eb99c337

https://x.com/AlkahestMu/status/1847516975767179397

@NathanpmYoung does this need to be like... verified or backed up in some way that it's actually just Claude 3.5 sonnet doing this, without human or other aid? Or would this resolve YES if repligate or some other user just releases a video they claim is of this?

Here's a video from maybe that same server: https://x.com/adonis_singh/status/1847707429066158546

This struck me as a little too good to be true when I saw it on twitter.

Not sure I'd call what I see in this video competent agents, and there seems to be some hand-holding from the creators, but these bots seem to manage to play the game okay: https://www.youtube.com/watch?v=1Sf437NKUPs

Still not clear to me how much is handled by the LLMs vs the other tools, since it seems that things like combat happen too fast for an LLM to react.

Title says "claude 3.5 opus" but the tweet is telling a story about sonnet being a competent Minecraft agent and opus just chatting. Is the title going to be fixed?

@MichaelEdgar Fixed