Issue 352: Just because you can, doesn't mean you should

AI has democratized much of our technical world, giving us access to skills and information without ever having to actually learn them. But just because you can do anything nowadays doesn’t necessarily mean you should. But we don’t have to go back to the stone ages (read: the pre-AI days). Sometimes, combining our traditional ways of working with our new ways of working is all we need to do to create really great software.

For example, we sit down with Quint’s Gabriela Moreira to explore why formal methods like TLA+ are essential for catching system design flaws before writing code, and contributor to the Rust language, Josh Triplett, chats with us about how Rust's strict compiler constraints provide the psychological safety needed for complex systems (and AI coding agents). On the blog, we’re exploring the build vs. buy trolley problem—and why building in-house isn’t always the right answer—and taking a long and hard look at the future of tech jobs with a Q&A with Andela’s Cory Hymel and a little self-reflection on how we hire at Stack Overflow.

Elsewhere on the web, we’re diving into whether NDAs cover AI tools, how much non-determinism actually matters in your codebases, and hhow likely those new system1 decision models are to pick saving humans over saving themselves in the trolley problem. In this case, just because you can read all of this in the links below, definitely means you should.