AI can write code, summarize incidents, generate Terraform, and sound incredibly confident while being completely wrong. For engineering teams, that creates a new kind of risk: the better the output sounds, the easier it is to stop thinking critically.

This talk is about what critical thinking looks like when AI becomes part of software delivery. I’ll break down where AI actually helps in platform, cloud, and DevOps work, where it tends to fail in dangerous ways, and how those failure modes show up in practice. That includes hallucinated infrastructure changes, shallow root cause analysis, bad assumptions hidden inside generated code, and overconfidence in summaries that skip key context.

I’ll also cover practical ways teams can use AI without outsourcing judgment: source verification, review loops, better prompting, sandboxing, policy guardrails, and clear ownership when AI is involved in production-facing work.

This is not a talk about whether AI is good or bad. It is a practitioner-focused look at how experienced engineers can use it productively without letting speed replace thinking. Attendees will leave with a framework they can use immediately to make AI-assisted engineering safer, sharper, and more useful.