Why LLMs Suck (aka the Software Developer's Guide to Keeping Our Jobs)

GenAI, agentic AI, vibe coding, Copilot, Claude Code, OpenClaw. We all know the hype and the buzzwords. Even your distant cousin who sends you videos of dancing cats knows. But more dangerously, your manager knows. And so the mass layoffs begin, even before the leftover employees have been trained to use these tools. You try explaining that the benefits in productivity and costs have not yet been reliably tested, but it is too late. Management only sees thousands of lines of code being generated in seconds, and that is what they care about. So what do you do? Easy: you come to this session, and share all the takeaways with them! Join us to learn about all the things that LLMs (and hence agents and any derived tools) are horrible at: from risks and challenges that are inherent to their architecture, critical issues that are not solved yet, to all the tasks where they perform pretty badly and we humans will (most probably) always be needed. All accompanied by personal learnings and experiences, but also major publicly known failures that costed companies millions. Millions that your company will save if they keep YOU in their workforce. Software developers: this is your quick-start guide on how to keep your jobs secure. No technical terms that managers will not understand, nor fluff that will make them miss the point. Simple, bold, direct facts that will open their eyes (backed with real business use cases and Excel-friendly numbers whenever a little extra scare is needed).