Hacker News Reader

Comment by antihipocrat | Hacker News Reader

Your performance on these tests would be equivalent to the highest performing model, and you would be much cheaper.

Investment in human talent augmented by AI is the future.

That’s the least reassuring phrasing I could imagine. If you’re betting on costs not reducing for compute then you’re almost always making the wrong bet.

If I listened to the naysayers back in the day I would have never entered the tech industry (offshoring etc). Yes, that does somewhat prove you're point given that those predictions were cost driven.

Having used AI extensively I don't feel my future is at risk at all, my work is enhanced not replaced.

I think you're missing the point. Offshoring (moving the job of, say, a Canadian engineer to an engineer from Belarus) has a one time cost drop, but you can't keep driving the cost down (paying the Belarus engineer less and less). If anything, the opposite is the case, since global integration means wages don't keep diverging.

The computing cost, on the other hand, is a continuous improvement. If (and it's a big if) a computer can do your job, we know the costs will keep getting lower year after year (maybe with diminishing returns, but this AI technology is pretty new so we're still seeing increasing returns)

The AI technology is new but the compute technology is not; we're getting close the physical limits of how small we can make things, so it's not clear to me at least how much more performance we can squeeze out of the same physical space, rather than scaling up which tends to make things more expensive not less.

Reply on news.ycombinator.com