Discussion
if you are trying to evaluate the efficacy of LLMs and you are reading the personal story of someone building an impressive demo, do not read the post looking for technical details. instead, see if you can spot:
- sleep disruptions
- grandiose generalized language about upcoming social and technological change
- increasing mania as the project progressed
- social isolation
- jokes about addiction that read as insecure deflection
- getting stuck at 90%
bonus round:
- baroque explanations of how the project needed to be restarted multiple times, and how this is really a learning experience about how LLMs work and a trail of false starts are evidence that things are actually improving so fast
the story of the mental health impact of agentic engineering is increasingly serious and science is not remotely keeping pace with the actual experience in the field. we have had a big problem in the software engineering community for decades with treating our teams and our *selves* as replaceable cogs who just output code and need to be able to white-knuckle our way through emotional impacts of the work and on the work, but LLMs are accelerating this tendency towards a true crisis
Install bonfire.cafe
Get the full app experience
