Hacker News Reader

Comment by donkeyd | Hacker News Reader

> I wager $50 five-ten years, that we'll be churning out content that an experienced artist is capable of.

I wager $50 that 'artists' will still exist and that this becomes a tool for those artists if they just accept it. We all have phones that can make great photos, yet I still paid for someone to photograph my wedding, because they are much, much better than anyone with a phone. Just understanding all the knobs of generative AI takes a lot of time.

I feel like the only thing actually impacted by this soon is the stock photography content farm game. Because if stock imagery is good enough, AI generated images might be good enough.

View on HN

"Replace" is difficult to define. Will there not be a single artist? Obviously wrong.

But possibly we have seen "peak" numbers in some professions? There are some exceptions, but in many modern professions, the numbers of professionals have only been increasing with GDP growth (e.g. quantity of lawyers/accountants/doctors).

I would guess there are more professional human translators today than there were in 1950. However, I think this is one profession that may "peak" at some point: For low-value texts, machine translation will be good enough, leading to a smaller pie of paid translation work, which can only sustain a smaller number of translators. The field will certainly stop growing and may decrease at some point. Anecdotal, but in the European Union institutions (https://www.politico.eu/article/translators-translation-euro...), it appears that we have already hit this "peak" and the future is gradual decline in the number of translators employed.

I did edit my post, but case still stands.

> Just understanding all the knobs of generative AI takes a lot of time.

We have more technological people in the field who can understand the knobs faster than we've ever had before. ML now has the ability to figure it's own knobs.

Who knew about Linux in 1994, compared to now? Then: not a lot, now? Half the internet.

Steam engines, very mechanical. Back then to drive a train, you needed to be trained, work with the conductor, driver. Now? A train driver operates a lever. Now AI are replacing those jobs for driver-less trains. Our local subway is exercising the idea.

ML, AI? Just search github and you'll find dime-a-dozen projects created from young-adults. It's only going to take a generation to burst this thing wide open.

I'm a 34 SysAdmin, it's past my time to jump in on the bandwagon. Not much I can do apart from feeding the servers that churn the cog.

> Yet I still paid for someone to photograph my wedding, because they are much, much better than anyone with a phone.

If you didn't have to pay and had some AI, ML do it for you producing the same results, would you?

The role of an artist will be replaced, they will co-exist. But the skill of actually producing the image will turn in to menial work; where for what their talent was once worth, skill learnt will suffer to be.

Anyone can type a "A fairy-elf of a dystopian world with a abandoned castle fighting off a horde of bionic-flies". All you would need is an artist to brush up on the artwork to make it feel more "human".

Maybe for the initial design, but once done. Why would you pay the artist, when you've already got the concept image required to be trained upon?

> But the skill of actually producing the image will turn in to menial work;

I don't see why it is a problem. Most people don't even give a shit about how much effort you put into the work. They only care about the end result. Spending a lot of time on something doesn't make it more valuable.

> Most people don't even give a shit about how much effort you put into the work. They only care about the end result. Spending a lot of time on something doesn't make it more valuable.

If you gave me a hammer and chisel, you won't get an end result, you'd get something that looks like junk.

Compared to someone who put's time and effort in to mastering the skill, you will get an end result.

Reply on news.ycombinator.com