[ldoughty]

Top picks:

Game development -- AI will eventually be guidable to write code, but it likely won't have "imagination" in the next decade (and I am not worried about my finances after that)

Beer making -- Similarly, AI likely won't understand taste for quite some time

Woodworking -- When everything is mass produced by automation, I suspect people will want unique, custom, hand-made things more.

Edit: Fix format, clarify statement on next decade :-)

[animuchan]

Game development is an interesting one. Looking at the bulk of recent triple-A titles, most of those feel so unimaginative and secondary that I can only assume AI already took over that field. :)

[the_only_law]

Funny enough though, the unimaginative, secondary nature that you describe has revived an old desire to make my own games, but I’m no artist.

[curioussavage]

Ha! So three very popular hobbies for programmers? I guess we’ll all all join you in those endeavors

[jobs_throwaway]

Programmers tend to have a lot of free time. You'd be hard-pressed to find a hobby that isn't popular among programmers

[throwaway675309]

Anecdotally but I find the following hobbies weirdly overly represented in IT:

1. Micro brew or craft brewing of some kind, Or at the very least an obsession with beer

2. Swing dancing

3. Frisbee golf

4. Cooking

As an engineer myself my hobbies don't intersect unfortunately:

1. Juggling and card manipulation

2. Sparta running training

3. Piano composition

4. Aviation

[jkmcf]

You forgot Whiske?y. One I don’t understand at all!

[bradly]

> Woodworking -- When everything is mass produced by automation, I suspect people will want unique, custom, hand-made things more.

Funnily enough I’ve used AI in my woodworking to generate novel coffee table designs, write product descriptions for Etsy listings, and generate a logo.

[rybosworld]

How did you generate a logo with AI?

[bradly]

I used DALL·E 2. I've had better success with it over Stable Diffusion. Prompt was something like "Create a vector logo for a woodworking site with a handplane and the words Burn Box Woodworking" For some reason it can never get the text correct, but it gave my a good enough logo to clean up in Photoshop.

[adamsmith143]

>Beer making -- Similarly, AI likely won't understand taste for quite some time

Might be wishful thinking, presumably you can give it taste reviews from people along with chemical makeup and can predict chemical mixtures that will rank highly on taste.

[talaba]

> beer making

No need for AI. T-test (https://en.wikipedia.org/wiki/Student%27s_t-test) was invented more than a century ago to improve beer quality.

[de_nied]

All your top picks have "innovation as a requirement for success" in common. AI is meant to reproduce outcomes, which means that innovation isn't something an AI can do. Or can it?

[more_corn]

Define innovation. AI can basically follow any prompt for writing or visual imagery.

If you can have the idea it can generate the output. You can probably prompt it to have ideas too.

[speedgoose]

> but it likely won't have "imagination" in the next decade

I find GPT models to perform very well on this task. ChatGPT is the best right now but even small models like GPT J 6B are good.

[ldoughty]

They work on text, but could ML/AI design SimCity 2000 in a way that is fun and enjoyable? or a game like Starcraft? (In the next 10-15 years, which is the scope of this Ask HN relative to me)

It might be able to make realistic text, even storylines, but I don't think ML/AI will be able to make GOOD games without significant guidance, especially things that 'break the mold' (Portal, Black & White, the first "Sim" game, the first RTS game)

[dogma1138]

ML/AI like every tool isn’t about absolute autonomy but rather about force multiplication.

This is what happened to farming, the textile industry, manufacturing etc.

It’s not that anyone thinks that an AI would be able to develop a game or an app from scratch but how many people it would displace by automating things from code generation to UI design and art assets creation.

We still have farmers, however their farms tend to be much larger and there are far fewer farmhands than there were 50 years ago let alone 100 or 150 before the Industrial Revolution kicked into high gear.

Ironically the productivity of AI/ML is quite likely going to open the door for many people the tech part of the tech community was quite often dismissive off - as in “I’m a big picture type of person and I can instinctively know how good looks like but I lack the knowledge to make my dream a reality”.

Whilst plenty of people that think they are like that often fall quite far from that even those who don’t like the Jobs of the world tend to be quite divisive individuals that too many people tend to dismiss as charlatans that failed upwards.

Overall I don’t think that an apocalypse is coming down at least not yet, however I don’t actually mind the fact that there might be more opportunity for people who really want to focus on the problem rather than people who just seem to know what the solution is or should be and find a problem that fits it.

And if AI can get to a spot where it can actually help people iterate solutions in a rapid and cheap manner it can really change how a lot of industries operate.

Also ironically a lot of other far less glorious IT professions are probably safe, old school sysadmins and infrastructure folks seem to be able to demand far higher wages than ever before which is what happens when 9 out of 10 new people in “IT” these days know how to package a hello world app that has 4GB of Node.JS dependencies and write the helm charts and TF deployment code which probably has more lines than the original Windows NT kernel but can’t calculate a subnet mask or know what MAC address is…

[cartoonfoxes]

If its furniture you're thinking of - people want custom hand-made things now, they just can't afford them. That trend isn't going to change with increased automation.