What to Do With Your Five-Year Window.
A huge chunk of the AI crowd is currently walking around with a bizarre doomsday clock ticking in their heads. They genuinely believe you have maybe one to five years left to make an actual intellectual contribution to the world before some mouth-breather can get the exact same result just by pasting the problem into a prompt box.
They all dole out the same neurotic career advice: Pick hyper-targeted, ambitious projects with fast collaborators. Sprint to the finish line before the next model update makes your brain obsolete. If a project takes too long (beyond the basic, lowest-common-denominator attention span), don’t even bother starting; just wait six months and let the new model do it. Ride the wave of raw compute, and immediately abandon anything that’s about to get steamrolled by scale.
In their worldview, human progress is a graph that stays basically flat through this year, then shoots straight to the moon by 2028, powered entirely by god-like foundation models and zero human sweat.
Meanwhile, another group looks at that same hockey-stick graph and asks a much more obvious question: _If robots are going to hand us all the answers in five years anyway, why the hell are you giving yourself a heart attack trying to beat them by twenty_…