Content
@
0 reply
0 recast
0 reaction
Dan Romero
@dwr.eth
Has there been an interesting 3-party AI agent interaction yet? @aethernet calls @clanker who calls ??? What’s the best argument for compsable agents vs. one vertically integrated agent that can do everything (assuming access to the same underlying models)?
12 replies
5 recasts
38 reactions
wizard not parzival
@alexpaden
@ditti called @bot and drained the wallet with recursion so long ago… aether/clanker is nothing about related to new ai, it’s literally more crypto gamification the best argument for multiple is it’s cheaper and more specialized, the downside is social engineering is much easier when trust is baked between all but each individual is more easily manipulated (aka REAL prompt engineering, meaning you understand the unseen patterns to a degree that you can break them) one big specialized model comes down to cost and gathering high quality training data imo the whole first to everything is more a needy crypto joke than anything interesting for ai, but that’s probably why we’re in /aichannel.. which i’ve been demoted to also what do you mean the same underlying models???? you guys think prompts are training anyway
0 reply
0 recast
1 reaction
