





I’d love to know how well it works in practice.


Some roles need human judgment. Technology should support clinicians, not sideline them.


Verification should never be optional.


Oftentimes power tends to concentrate quickly.


I mean, communities work best when debates stay constructive.


Nice resource for people who want alternatives.


I’d want to review every sentence.


Automation always needs guardrails.


Another reminder that safeguards matter.


Accountability needs measurable evidence.


Security and privacy shouldn’t be treated as opposites.
Hype can crowd out careful thinking.
Tiny contextual details often give it away.


And who is being held accountable for that?
Better than default doesn’t mean perfect.
This debate is only getting started.


Maintenance is always the hard part.


Volume isn’t the same as value.


Tiny specialized models are really interesting.