LAIer 8|9 - Noch'n KI-Podcast

LAIer 8|9 is a brisk German AI show for people who are already surrounded by AI talk and want the noise sorted into something usable. It is not gentle hype. The conversations begin with the kind of irritation many users recognize: a favorite model suddenly feels worse, token costs creep upward, a context window gets treated like an infinite fuel tank, or a local AI setup promises savings while quietly demanding hardware, tooling, and patience. From there, the show widens the frame. Short complaints become arguments about model design, prompting discipline, workflow architecture, enterprise software, security habits, copyright law, and political responsibility. The pacing is loose but not sloppy. Barbara and Jochen bounce between naive questions, technical boundaries, business examples, and dry side remarks, leaving enough live thinking in the room for the listener to hear the distinctions being built. The show is especially good at separating productivity from value. It keeps asking whether faster output actually produces better work, or merely hides weaker judgment behind automation. That same skepticism carries into copyright and AI training, where it resists easy slogans about theft or democratization. Military and government use get a more grave treatment, especially around the Pentagon, GovCloud, Palantir, ICE data, and mass surveillance concerns. Cybersecurity episodes stay practical, linking AI agents and penetration testing to very ordinary failures that still hurt organizations. Even when the topic is local AI, the promise is measured against real constraints: command-line comfort, old devices, sensitive data, and what happens when a person experiments without a safety margin. LAIer 8|9 works because it does not pretend competence is a vibe. It treats AI literacy as a messy, adult skill: part vocabulary, part cost control, part technical hygiene, and part knowing when the shiny object is making the user poorer rather than smarter.