We’ll explore the subtle yet significant ways in which reliance on AI tools—particularly large language models (LLMs) can change the way developers think. Drawing from personal experiences and broader industry trends, the talk examines the dangers of vibe coding: an intuition-driven, unstructured approach that LLMs often encourage, where solutions are stitched together based on plausibility rather than understanding. We'll consider how over-reliance on these tools leads to cognitive offloading, potentially dulling critical thinking, weakening foundational skills, and diluting the craftsmanship that once defined great code. Rather than rejecting AI assistants outright, this talk invites a more mindful, deliberate engagement, where tools support, but do not substitute, human judgment and creativity. The goal is to challenge developers to rethink their relationship with AI, reclaim ownership of their cognitive processes, and rediscover the joy of making things well, not just fast.