Leave Me Behind: An Android Dev's Case Against AI-Assisted Coding

A veteran Android developer pushes back on the ultimatum that engineers must adopt LLM-based coding tools or be rendered obsolete. Tracing a decade-long career that began with a college Java class and grew through hackathons, conferences, and mentorship, he argues that what made the craft meaningful was never the output but the human network around it — the office mate who patiently explained RxJava, the Stack Overflow strangers who challenged his assumptions, the Droidcon speakers who taught for free.

After giving modern AI coding assistants a fair trial, he found that even when they produced working code, they hollowed out the parts of the job that built durable skill: the trial-and-error of testing architectures, the trade-off debates with colleagues, the slow comprehension that comes from researching a problem rather than accepting the first generated answer. Delegating critical thinking to a prediction machine, he warns, erodes the very abilities engineers need to build resilient software.

The broader concern is cultural. LLMs are trained on a generation of developers who learned and shared in public, and routing every question through a black box threatens to dry up the open knowledge-sharing pipeline that trained those models in the first place. His response to the ‘adapt or die’ framing is simply to opt out and keep coding the way he learned.