AI Agents Speed Up Replacement of SWE and MLE Roles in Frontier Labs

That said, it's a funny issue. Do AI agents render human intelligence or human knowledge *more* irrelevant? We're doing better on reasoning than on continual learning now; stale knowledge cutoffs give the edge to people who've published recently and basically know the hottest tricks and keywords. But if you reduce the function of a researcher to prompt engineering, isn't it a fairly desperate endgame? Your hiring pipeline gets reduced to cope with the next generation of search agents, such as the one @victor207755822 is building. I would still bet on human ingenuity as the more durable good.