Ask HN: How are you planning for a post-knowledge-work world? | Hacker News Reader

Ask HN: How are you planning for a post-knowledge-work world?

3 pointskypro3 comments

I struggle to see how the majority of digital knowledge-work jobs will survive the continued progress AI over the next several years.

Even if you believe there's a chance not all digital knowledge-work jobs (coders, graphic designers, copy writers, etc) will go away, there's a good enough chance of this happening that you should probably still prepare just in case – especially if you have dependants.

Just wondering what preparations people here are making? Ideally I'm looking for career ideas for an ex-coder who's never done anything but IT work.

View on HN

"Why, we're all seaman here, I should think."

"We're all foc's'le hands, you mean. We can steer a course, but who's to set one? That's what all you gentlemen split on, first and last."

(From Treasure Island, by Robert Louis Stevenson)

Say an AI can write the code. Can it decide what the code ought to do? Can it look around the world (or around the business), see the need, and write the specification? Once it writes the code, can it debug it? Can it maintain the code over the next two decades as the world changes around it?

If an AI can take a spec and write the code unsupervised (which, personally, I still doubt), all that means is that we get to level up. We now figure out what the code needs to do and write the spec.

You say "but that won't take as many people!" Maybe not. Or maybe we keep the same people and write more software. It's not like the people we have were writing all the software that people wanted written...

Indeed. And it's not like the software we have is all that good! A lot of AI effort could go into making software that people like better, that's easier to use, and easier to maintain.

The real "knowledge work" is applying the human knowledge of one's own experience and education to solve a presently-unsolved problem, and to decide to do in the first place.

Learn as much as you can about different industries and challenges they are facing.

If AI can write code to specification, then the person with the advantage is the guy who knows an actual industry problem that people will pay money for. The majority of industries, upon close inspection, are technologically backward major ways, but previously it used to be really hard to bridge the gap for a person that actually knew the problems. With A.I it should be way easier.

For example, in finance/trading there is a lot of software that is old, inefficient but just passable enough that no one will try getting an MVP out. Perfect example is POEMS(Portfolio/Order/Execution Management System(s)), i.e what you trade with, if you spend any time researching this you will realise that a lot of the products in use are super disruptable but MVPs are hard to build.

Reply on news.ycombinator.com