95b37377
ive been using copilot and claude code daily for 6 months. here's what actually happens:
1. it writes boilerplate faster. great. that was the easy part anyway
2. it makes subtle bugs that take longer to debug than if i wrote it myself
3. it has no idea about our internal systems, infra, or business context
4. it cant attend a meeting, understand a vague product spec, push back on a PM, or navigate team politics
the 10x engineers using AI are the same people who were 10x before AI. the mediocre engineers using AI are now generating mediocre code faster.
stop panicking.
65fe1dd7
this is cope. AI today cant replace you. AI in 2 years? 5 years? you have no idea and neither do I. the trajectory is clear even if the timeline isnt
4a6dc182
agree with OP. the bottleneck was never typing code. it was understanding wtf the business actually needs and translating that into systems. AI doesnt do that
2a7ce999
"it cant attend a meeting" is doing a lot of heavy lifting in this argument. most of my job is coding not meetings. if AI handles 70% of the coding part, companies need fewer of me
ddbd4252
the correct take is: AI wont replace SWEs but it will reduce the number of SWEs needed. we went from needing 10 devs to needing 6 to ship the same product. that means 4 people are looking for jobs. thats the real problem
82a800e6
this exact post was made about stack overflow in 2010, about google in 2000, about IDEs in 1990. the tools get better, the bar rises, and we build more complex things. the number of SWEs has only ever gone up
79b2b213
past performance is not indicative of future results. this time the tool literally writes the output. previous tools just made you faster at writing it yourself. thats a categorical difference
1a4f7f3a
both sides have good points. the truth is nobody knows. hedge your career by being good at the parts AI cant do: system design, stakeholder management, ambiguity
