To your point, having AI-assisted tools to build apps (with React Native or anything else) is part of your experience IMO and interviews should account for that.
At Expo I tell the team using AI is a skill and in 5 years we will be glad to have been practicing it day by day (or regretful and possibly feeling threatened if we do not). We are also embracing AI in interviews for this reason.
However it can be useful to show or tell an interviewer about an actual RN app or native module you built. Having run into the “unknown unknowns” is valuable especially with how inconsistently (un)reliable LLMs are. It’s worth going through the motions at least once.
On this tangent we are working on dependable AI, most recently with Expo MCP Server which feeds up-to-date docs to LLMs and lets agents interact with mobile simulators (tap, screenshot) and verify their work before interrupting the human. In the future I hope we can feed app state, the view hierarchy, and all types of errors to the agent.