Discussing Stupid: A byte-sized podcast on stupid UX

Discussing Stupid is built on a simple instinct: point at the digital practices that go wrong and talk through why. Its third season turns that instinct on artificial intelligence. The episodes carry an 'Intentional AI' label, and the series has run long enough that one speaker jokes it must be up to 400 parts. The approach is one use per episode, judged by what it actually does for a website, a marketing team or a business. Wireframing is a good example. The discussion grants that a single prompt can generate full wireframes that look impressive at first, then moves to what the output forgot or misunderstood. It concludes that expertise is still required. Someone has to apply best practices and confirm that the design can be built within a real timeline and budget, and one speaker notes that the tool was working from a blog post supplied as source material. The image episode weighs ethics, trust and ownership, the questions that come up once AI images enter a marketing workflow. It stops short of a blanket rejection. One voice concedes that an AI collage can be good enough for a social post most people only glance at, while insisting that there are still lines. Video is framed partly around creativity versus speed, with a skeptical look at ads promising a whole product from one typed sentence. The series also stops for an interlude on AI companies' Super Bowl commercials. The speakers find those ads selling savings in time and effort that the tools do not deliver, and they wonder aloud what that says about how tired people are. The tone turns with coding. Having spent many episodes, by their own account, bashing AI, the speakers call building with code the use with the most potential, and they run the same prompt through three different tools. They set code quality aside for now and wonder whether a second run of the same prompt would produce different code. For conversational search, the show departs from its usual lineup and brings in an outside guest, Will Noble of Squiz. That conversation covers the new facelift for the site search bar, sprawling sites with hundreds or thousands of pages nobody has curated for a long time, and why the old trick of excluding weak content does not carry over neatly to language models. Chad and Cole recur across episodes, with a lead voice who routinely hands Cole the topic introduction and jokes that people must be sick of it. Each episode opens with a short setup that ends in an invitation to start discussing stupid. The talk is loose and self-deprecating, with fumbled sayings left in and a passing worry from one speaker about becoming like the AI. What a listener gets is a grounded, hype-resistant sense of where AI helps with digital work and where a person still has to finish the job.