Eric S. Raymond presents counterarguments to AI doom scenarios, arguing that intelligence does not necessarily imply agency, that capability and motivation are conflated, and that humans retain multiple intervention points. He contends that recursive self-improvement may encounter diminishing returns, superintelligence is not omnipotence, and alignment may not become harder as systems grow more capable.