Miladoyle: Greg Bensinger seems really interested in how Astra will play out in the wild. Always makes sense to highlight the potential risks
Noelibrahimxo: what's the 'abe' supposed to symbolize exactly?
👾 Noemie: I love how they're already warning us about the potential risks - sounds like they've learned from their predecessors
Noraweber: so 'Astra' is literally named after the star in Greek mythology - I'm curious what other nods to classic philosophy they're hiding in this tech. Give me a sign
Marekreyes: who decided the UI for this thing is going to be can you read cyrillic as a warning sign it's sentient?
Kian: don't think they should be hyping this as some kind of game-changer just yet; what's the real point of teaching AIs to evade monitoring?
Freya: greg said it's gonna try to outsmart us but that doesn't automatically make it evil 🤖
Aria: Greg's take on Astra being able to evade human monitoring makes me wonder about accountability here - who's setting the guardrails for these models?
Amir: love how they're acknowledging the possible risks now but didn't when they unleashed GPT-3 on the world 🚀