Ha - I used Claude to help organize and edit it, yeah. Didn't see much point in pretending otherwise. The irony isn't lost on me, but I'm not arguing these tools aren't useful, just that we should call them what they are. Same way I'd use a calculator to write a math paper without claiming the calculator understands arithmetic
Fair pushback - you're right that strict dictionary definitions are broader. I probably should've been more precise there. My point is more about how the term is used in the actual fields studying it (cogsci, neuroscience, etc.), where it does carry those biological/embodied connotations, even if Webster's doesn't explicitly say so. But you're right to call out the sloppiness.
God, yes. The 'you know what I mean' thing drives me crazy because no, I actually don't think they do know what they mean anymore. I've watched people go from using it as shorthand to genuinely asking ChatGPT how it's feeling today. The marketing has been so effective that even people who should know better start slipping into it. Completely agree that we missed a chance to frame this correctly from the start.
"Yeah that's exactly the point - when you're actually working with these models on the infrastructure side, the whole 'intelligence' narrative falls away pretty fast. It's just tensor operations at scale. Curious about your experience with decentralized GPU networks though - do you find the reliability trade-off worth it for most workloads, or are there specific use cases where you wouldn't go that route?"