84% of devs use AI. Only 33% trust it. Why are we pretending this is fine?

Jason Baum, Project Leadership at Selenium, is here to say what most won’t: AI is creating as many problems as it solves. Development is faster, sure, but the cost is rising. More bugs. More hallucinations, too. Teams are losing time chasing the promise of AI without fixing the foundation it runs on.

In this episode, Jason and Nicky get real about the AI productivity paradox—why developers are feeling slower and what teams can do to get quality back on track without tossing the whole tech out the window.

If you’ve been asking yourself, “Am I actually building faster or just debugging more?” this is the reset you need.

In this episode, you’ll learn:

1. Why senior devs are the most skeptical AI users
2. How AI is shifting the definition of “quality” in code
3. What it takes to turn AI into a teammate instead of a time sink

Things to listen for:

(00:00) Meet Jason Baum

(01:19) Developers are using AI they don’t fully trust

(03:15) How hype outpaced AI’s actual capabilities

(05:59) The AI calculator analogy: helpful, but dangerous without skill

(08:47) Is AI improving or breaking software quality?

(11:03) The real purpose of “vibe coding” tools

(13:15) Why senior developers are wary of AI’s output

(18:45) Critical context: What AI still can’t grasp in code

(22:45) The rise of citizen developers using AI tools

(25:05) Risks and realities of using MCPs with LLMs

(31:20) Can AI-driven testing really catch what matters in code?

(34:08) Why AI is better at analyzing tests than writing them

(39:25) Why most AI projects fail before they even start

(45:12) Preserving developer intuition and problem-solving in the AI era

(51:02) Final thoughts: where developer-AI collaboration goes next