The AI Safety Crisis: Are We Ready for Superintelligence? | AI Safety Expert Roman Yampolskiy | SparX

In this episode of SparX, we speak to Roman Yampolskiy, a leading AI safety researcher and professor of computer science, about the risks of creating superintelligence and whether humanity is prepared for what may come next. Roman argues that we may be closer to human-level AI than many assume, and that permanently controlling a system more intelligent than humans could prove fundamentally impossible. He lays out why some researchers believe development should slow down, and why the window for meaningful intervention may be narrowing.

Roman discusses the rapid acceleration toward AGI, early signals of job displacement that could become visible by the end of this decade, and why traditional patterns of technological disruption may not apply this time. He explains why large companies continue investing heavily in AI despite debates around scaling limits, how the global race toward superintelligence is unfolding, and why no scalable safety mechanism currently guarantees control. The conversation also explores AI consciousness, digital labor, the simulation hypothesis, and what widespread automation could mean for identity, purpose, and humanity’s long-term future.

If you’re looking for a rigorous and research-driven perspective on the technical, economic, and existential implications of advanced AI, this episode offers a serious examination of what the next decade could hold.