The current AI boom is being compared to the dot-com bubble, with differing perspectives on its transformative potential. While some, like Andrej Karpathy, see AI integrations as revolutionary "new paradigms," others view them more pragmatically, as mere "Slack integrations."

This divergence mirrors the dot-com era, where technologies were simultaneously "wildly overhyped" and genuinely "changes everything." During the dot-com crash of 2000-2002, Amazon, which lost over 90% of its stock value, ultimately succeeded by understanding its core capabilities, unlike Pets.com which failed. The key was asking the right questions about what the technology could uniquely enable.

Today, teams are grappling with how to integrate AI, often shipping code before fully understanding it. The debate isn't just about model quality but about fundamental workflow changes. Karpathy hypothesizes that models won't become dependable enough soon to settle these questions, suggesting the need for human understanding and careful judgment to navigate this era, much like in 2000.