How long do you think? I give it 3 years | Hacker News Reader

How long do you think? I give it 3 years

6 pointshmokiguess9 comments

Who’s ready here? I feel like in 3 years time we are being forced out into retirement.

I already replaced the junior devs I work with entirely, soon I’ll be replaced.

A few years back actually, before AI as we know it, there were lots of quants in finance making millions who now are algo bots that cost in the hundreds of thousands but still lot less than millions in commission (plus no sick day, etc, etc).

You get the point, I’m not asking “if” I’m asking “when”. What’s your take on the question? I give it 3 years.

View on HN

LLMS still can’t troubleshoot. I have been thinking about how to solve it and I think an entirely new architecture is needed. Quants got replaced by quant bots the same way human calculators got replaced by electronic adding machines.

I haven’t met an AI who can think clearly in a systems setting. I’m guessing it might be a decade or two. As an example, ask an LLM to do some 10th grade math. Inspect the thinking process. It can regurgitate the process and the rules but cannot perform them.

Same with troubleshooting.

>As an example, ask an LLM to do some 10th grade math. Inspect the thinking process. It can regurgitate the process and the rules but cannot perform them.

It seems to me that the solution is just RL to get the language model to delegate the actual calculation to the appropriate tool.

I'm not sure your explanation is correct. The quant bots are also decision systems who attempt to predict in non-deterministic environments.

The reason they got replaced isn't because the problem became deterministic (like a calculator). It's because the error rate and the cost got to an acceptable place when compared with the cost of a human quant.

I think it is never: while we should be able to achieve more with less, smart people show up who can sell more stuff and everyone keeps their job?

I'll go with <3

that soon eh

No later than 2030.

Companies cut ~600K tech jobs since covid purely on AI hype. Now there is actually (some) utility. The bubble will burst at some point (they always do). I think companies cut to the bone at that point. No one knows when the top is but when it turns I don't think many of the cushy knowledge worker (white collar) jobs come back. It was a good run. 1975 - 202X?

This seems like a baffling comment. If AI is good enough to replace the vast majority of knowledge workers, why would the AI bubble burst? (Or, more accurately, why would you describe it as a bubble?)

Reply on news.ycombinator.com