Every developer survey now says the same two things, and they look like a contradiction. Adoption of AI tools keeps climbing, with most developers using them or planning to. At the same time, trust in what those tools produce keeps falling, and the gap between how fast developers feel and how fast they measurably are shows up again and again in the research. That is not a contradiction. It is what a maturing relationship with a fast, fallible tool looks like, and it is the real story of AI-assisted development in 2026.

In this data-driven session, Brian maps where AI-assisted development actually stands, using defensible numbers from primary research rather than vendor marketing. You will see who is using what, where AI genuinely accelerates work, and where the "almost right, but not quite" problem quietly turns time saved into time spent debugging. Brian will show the verification gap up close: how much code AI now writes, how little of it gets fully reviewed, and why reading and validating code is becoming the core developer skill. The economics matter too. As frontier cloud models move to usage-based pricing, the cost of AI-assisted work is getting harder to predict, and more teams are weighing self-hosted and local options against it, a tradeoff Brian explores in depth in a separate session this week. Live demos ground the numbers so you see the wins and failure modes for yourself. You will leave with an accurate, current read on what is real, what is hype, and where the profession is heading, with the evidence to position yourself and your team deliberately rather than by reflex.

You will learn:

• Where AI adoption, trust, and measured productivity actually stand in 2026, drawn from primary research rather than vendor claims
• Why the "almost right, but not quite" problem and the verification gap matter more than raw acceptance rates, shown with live demos of both
• How the developer role is shifting from writing code toward validating it, and what that means for the skills worth investing in next