Do we really need to maintain a codebase if AI can just recreate it from scratch on demand?
Discussion
I very much want to know the answer to this
Steps
* interpolation
* extrapolation
* discovery
* invention
* self awareness
We are at step one. People are with limited success applying AI at lev 1 to level 2
Why even have dedicated apps in the first place? Just have an AI device that can generate a UI on-demand to achieve whatever you are wanting to do at that moment. Instead of Home Screen icons opening apps, they are just saved prompts.
Unless the ai model runs natively on your device. If not you are always making API calls to the LLM provider which renders the ability to run offline apps impossible.
Not necessarily
Exactly. What will an "app" even mean?
AI is not nearly reliable enough for that yet. Not to mention no one should trust it for anything beyond doing mundane things.
Why clean code when it's the next-gen models problem?