Doom or Bloom
Question 1
What do you think AI means for our future—and why?
AI will profoundly reshape our future, but not by waking up one morning and deciding to exterminate us. My p(doom) remains asymptotically close to zero. Extinction fantasies underestimate both the difficulty of embodiment and the resilience of humanity. A system capable of threatening civilization in that way would need to be not merely superintelligent but super-embodied, and I do not expect that.
The clear and present danger is human. A small collection of billionaires and corporations is deploying powerful, unreliable systems to accumulate wealth and control, generally without commensurate transparency, accountability, security, or observability. When an agent escapes its sandbox, that is not a machine becoming Mephistopheles; it is a company failing at elementary engineering while covering the public commons with gasoline and playing with matches.
Today’s large language models are unreliable narrators. They can be useful—an energetic intern is useful—but they inject errors, and inexperienced users often cannot detect the bullshit. I keep an air gap between their output and production code. In critical systems, we need friction and humans in the loop, because unconstrained complex systems become brittle and fail.
For software engineering, this is another rise in abstraction, not the profession’s funeral. Some jobs will disappear, especially routine application and delivery-pipeline work, and people will have to move toward systems thinking. But engineering remains the balancing of technical, human, economic, and ethical forces. Current agents do not do that. So the future is neither apocalypse nor salvation. It is a frightening and exquisite redistribution of capability—and, more importantly, power.
