Sep 16 5 min
Jev means structured output is interesting again
Sep 15 5 min
Tell agents the why, not just the how
Sep 14 5 min
Slow developer experience will bottleneck fast models
Sep 13 5 min
AI is breaking our proxies for expertise
Sep 12 5 min
Don't build tools for AI agents
Sep 10 5 min
They really do think AI might kill everyone
Sep 09 5 min
Why we should anthropomorphize AI agents
Sep 08 5 min
Automatically detecting AI text in my browser
Sep 04 5 min
Radical responsibility means treating people like tools
Sep 02 5 min
How to protect yourself from workslop
Aug 30 5 min
You have to beat the models at something
Aug 22 5 min
You should never be angry at work
Aug 21 5 min
Readers can't identify watermarked AI text
Aug 19 5 min
Good writing is obvious, not original
Aug 16 5 min
AI text watermarking is not a big deal
Aug 11 5 min
No, local models will not win
Aug 09 5 min
I got an email about resistance
Aug 02 5 min
Giving and taking credit in big tech companies
Jul 31 5 min
AI models need moral support to make discoveries
Jul 29 5 min
You don't have to be smart if you can think clearly
Jul 23 5 min
Powerful AIs might escape containment by releasing themselves as open-weight models
Jul 19 5 min
Impro is a handbook for running a cult
Jul 18 5 min
Overtraining as the path to human-like AI
Jul 14 5 min
What does "playing politics" mean for software engineers?
Jul 11 5 min
In defense of not understanding your codebase