Ask HN: Will it be better to be a specialist or generalist in the age of AI?

Currently I am unemployed SE, and seeking a job, but also very wary of the changes occurring in the space. I know the field won't be recognizable in 5 years from what it is now, so I'm looking to get myself back on my feet while also preparing optimally for the coming changes in the landscape, and how that bodes for certain ranges of skill.

One one side, most AI LLMs seem superhuman on easy entry level programming tasks, but increasingly falter as the complexity increases. This seems to indicate those best shielded against AI will be the very niche specialists capable of solving very hard, specific problems

On the other hand, AI tends to perform better when tasked to solve more specific problems, and more general, abstract problems that humans are better at tend to be harder to crack. This seems to indicate a generalist approach will offer humans a greater competitive advantage.

Which makes more sense? And of course, if the advent of AGI were to occur in that timeframe, either aim would be irrelevant.