Remix.run Logo
▲ sandeepkd a day ago

I am more than convinced that the folks who understand how LLMs work, what they are good at and what their limitation are would make it out on top. Anyone who is still hands on would do just fine.

Within couple years there is definitely going to be a shortage of people who understand how systems work. The only black swan possibility is the enterprise industry getting too small cause of heavy consolidation.

▲sterlind a day ago | parent | next [-]

within a couple of years I expect LLMs to understand how systems work well enough, and to be autonomous enough, that deep understanding of systems becomes largely irrelevant. right now LLMs can't manage the technical debt they create, but that's a problem that can and will be RL'd away. it probably hasn't already purely because the trajectories to sample are uneconomical to generate.

of course, I'll still be doing systems. maybe from my cabin in the woods, hiding from the machines, but I'm in it for the love of the game.

▲sandeepkd a day ago | parent [-]

I highly doubt that, at the same time thats always a possibility, going in the direction of machines ruling the world

▲mycall 20 hours ago | parent | prev [-]

Systems might become even more opaque and morph in the next few years as AI models begin to make their own stacks for solution designs. I for one would welcome simplifying the overly complex stacks upon stacks of abstractions we have today. Instead, move back to having well-formed, proved unikernel blobs of execution (e.g. Symbolics). The sky is the limit.