Interesting article! However, I’m curious about your stance on AI’s inability to interact with local processes. For example, I recently used Google’s Gemini CLI to start a Rails server on port 3000, but since I already had a server running on that port, Gemini detected the conflict and notified me about the existing process and its PID.
Doesn’t this imply that AI tools are already capable of interacting with and reading from the system environment (like ports and running processes)? How does this fit with your point that AI assistants today can’t access or control background processes? Would love to hear your take on this apparent contradiction.
Interesting article! However, I’m curious about your stance on AI’s inability to interact with local processes. For example, I recently used Google’s Gemini CLI to start a Rails server on port 3000, but since I already had a server running on that port, Gemini detected the conflict and notified me about the existing process and its PID.
Doesn’t this imply that AI tools are already capable of interacting with and reading from the system environment (like ports and running processes)? How does this fit with your point that AI assistants today can’t access or control background processes? Would love to hear your take on this apparent contradiction.