Self-Improving AI and Loss of Control Risks: A Plain-English Intro

This is a free, 90-minute online workshop from AI Safety Awareness Group Raleigh on self-improving AI and the risk of losing control of it, aimed at people with no technical background. It begins at 8:00 pm on November 18, 2026.

AI companies already use their own systems to help write code and speed up research. The session asks what could follow if AI became capable of building stronger AI, a process researchers call "recursive self-improvement." It is run as an interactive workshop rather than a lecture, and it is open to newcomers, the casually curious, and anyone unsure what to believe about AI.

Topics covered

• Self-improvement, and why some researchers expect progress could accelerate sharply once AI helps build AI.
• Smarter-than-human AI, and how it would differ from the systems in use today.
• Power and control, including the economic, political and strategic influence advanced AI might hold, and why "just turning it off" may not be realistic.
• Misalignment, where a system pursuing the wrong objectives causes harm without any malicious intent.
• Societal choices, meaning how decisions by researchers, companies and governments may shape whether AI stays under human control.

What participants get

• A non-technical introduction to core ideas from AI safety research.
• Live demonstrations of current AI capabilities.
• Small-group discussions and thought experiments linking AI risks to everyday life, institutions and incentives.
• A better mental model for judging headlines, hype and claims about AI progress.