Hacker News Reader

I’ve been slowly warming up to this idea, and I’m not sure I’m totally convinced, but I think it does make sense to compare the web development industry to the art industry here.

Have web builders totally replaced developers? No, not at all. It’s definitely cut out the lowest end of the market, but that’s because your local restaurant really doesn’t need more than a squarespace template.

But warping a website builder to do something more complex is complicated. To the point where you need to hire someone trained in using that builder tool. Then that person finds the edges of the customizability of that tool and reaches for just writing HTML/CSS/JS.

I could totally see a similar trajectory here. We’ll have prompt engineers, and they’ll realize sometimes photoshop is just easier to tweak the output, then maybe they’ll realize it’s easier to prompt for components of an image, and composite them manually… and then we’ve reached a point where it’s just a tool that makes art easier.

Though… it still does feel somehow different from web development. Will just have to see how things shake out.

I wish this were the case but I'm not really sure I buy it. I'm a software engineer with absolutely no eye for design or illustration. Now I can write a prompt to get me a good-enough image in 90% of cases whereas before that was totally unreachable without an artist. I don't need any training to use this tool and in many cases it gets me to the full end result without any modifications needed.

The reverse doesn't hold -- there are no tools that allow an artist with no training or study to be an effective developer. I'd say the CMS example is more of a data entry job than a development job now, it's two different things.

Most of the advancements in tech still need trained developers to utilize them. GPT-4 cannot create non-trivial programs itself and does not seem particularly close to doing so, it's mainly used for scaffolding and bug fixing and guiding, all of which still need a trained developer at the wheel.

I'm worried for the future of artists.

Were you using artists for anything you can use DALLE for now? Probably not, because what they're best at is too different. But making them more similar will actually increase demand for them because you'll have higher standards for whatever you get out of them. (Jevon's paradox)

Basically, there is no circumstance under which automation reduces demand for labor. It's like if you got a pay raise and started worrying it might make you poorer.

> Were you using artists for anything you can use DALLE for now?

I mean sure yeah, in the past I would pay for logos from artists and now I just make them with AI for my projects instead.

> I'd say the CMS example is more of a data entry job than a development job now, it's two different things.

Right, but it used to be a development job, which is the point. The CMS took a development job and turned it into a data entry job -- exactly what DALL-E does to some limited forms of artistry.

But development didn't disappear, and neither will artistry.

That's true, but I still worry because it feels like we are reaching some kind of inflection point. For instance, Vercel's latest product (https://v0.dev) makes me nervous about the impact to front-end development.

What's interesting about seeing these conversations on Hacker News is that 10 minutes ago, our profession was putting thousands out of work on a daily basis without even a second thought.

Ever stopped at one of those self-checkout registers at the grocery store that didn't exist in your childhood and wondered what happened to the cashier who used to be there?

A lot of people mumble something about "jobs not worth doing", but I'm not sure if web front-end development (how many unique, differentiated, worthwhile websites exist? how much of that code is glue and boilerplate, and people solving the same problem in parallel every day?) fares much better in that gauntlet.

Software developers right now are, for the first time in their lives, experiencing a little backlash to their privilege -- for hundreds of millions, they've been riding that rollercoaster all their lives.

> Ever stopped at one of those self-checkout registers at the grocery store that didn't exist in your childhood and wondered what happened to the cashier who used to be there?

They're still there, they're restocking shelves.

Similarly, the invention of ATM machines increased the number of bank tellers.

My local supermarket normally seems to be staffed by maybe two or three people total. They stock shelves until someone needs them to fix the machines, do an age check or temporarily use the old checkout line. I remember supermarkets of that size used to have several people stacking shelves and then the same number again on the tills. So I think it has reduced the number of those jobs.

I feel embarrassed to admit that I've never thought about it from that perspective. Totally agree.

You're right, and I'll gladly bite that bullet and say that the changes you describe are good. It's almost always a net positive to be able to produce more output with less human effort.

There are two different questions:

- will be more artwork be bought (probably)

- will the increase in the quantity of artwork being bought (assuming there is one) compensate the decrease in price for each piece of art?

Clearly for software, _so far_, the increase in productivity / decrease in price has led to such a big increase in demand for software that some programmers are better off - I say _some_ because many jobs within the software industry stopped existing.

Yet if we look at something like agriculture, there has been some increase in demand in food products(e.g. much more meat is being sold), but one can only eat so much - so most of the increase in productivity has led to way fewer people working as farmers, and not much increase in farmers income.

No "these are tools to replace farmers - these are tools farmers can use to do more and better work".

Basically people are making statements about the elasticity of demand with respect to prices for prices & quantities no one has ever observed. If making a piece of art is now 1/2 (random number) the cost thanks to AI, will people buy 50% more? 100% more? 1000% more? I have no idea and I am not sure why people think they do.

This certainly seems to reduce how much experience and study an artist needs to have in order to produce art. And I think it's likely to reduce how much the viewer values the product as well.

Automation does not always increase demand for the jobs being automated. Tractors didn't create more demand for oxen than ever.

The automation of tractors over manual plowing didn’t replace the Oxen, but the people driving them.

Farm automation allowed a single man to plow many fields in the time one field would have taken.

Automation allows one person to accomplish more with less.

Tractors didn’t reduce the demand for product, but increased it many fold.

Automation increases demand; every time.

The number of people working in agriculture was divided by c. 6 in the US [0], despite the population being multiplied by c. 3.5.

Sure, automation increased demand for food, but demand did not increase enough to not destroy 5/6th of farmers jobs.

Now this may not be a bad thing at all, that's not the point, but I am not sure "hey demand for art will increase but 5 out of 6 artist jobs will no longer exist!" is a super hopeful statement to artists.

[0] https://ourworldindata.org/grapher/number-of-people-employed...

> Automation increases demand

Yours is a different definition of demand than the economics definition, which is basically the amount customers will buy as a function of price. Higher demand means buyers will pay a higher price for the same art.

Automation increases supply, which increases the volume of art consumed, mainly because it is available at a lower price. That is shifting to a different point on the same demand curve, and doesn't mean that demand has increased.

Automation increases demand.

Just look at the dairy industry. Pre-automation sales numbers vs. post-automation sales numbers.

This is true for every industry on earth, and doesn't care about the pedantic definition of demand. Consumer demand = more purchases. More purchases = more revenue.

What's the alternative? Outlaw AI art? Too late for that.

It's inevitable, but that doesn't mean we should delude ourselves into thinking it's win-win for everybody.

Who's deluding themselves?

To refresh your memory, the context of this thread is that JimDabell argued that software automation increased demand for software programmers, and the same will be true for art. I responded that there's no guarantee of that, and then you pointed out that it can't be stopped. Which may be the case but it has no bearing on the question of whether artists will lose their careers. Just because it's inevitable doesn't mean it's not a cause for legitimate concern.

But despite the work, it still takes a programmer to glue all of our crap together even if that crap is mostly preassembled. That's our fault and frankly a failing of our industry but it's why automating most of programming doesn't mean that your team can get by with 0 programmers.

Clip art/stock photography though, you can do that with 0 artists if you have the actual art bit automated.

>you can do that with 0 artists

you could do that with zero artists for decades because it's already fully commoditized. Everyone who is cheap already just buys a subscription to some gigantic vault of stock imagery owned by some digital rights holder for pennies.

the mechanical reproduction of art did not start in 2020

It's a good point, I'm not a consumer of that market so I don't know it well and am projecting my lack of even rudimentary mspaint skills that in that situation. I wouldn't be able to remotely customise something I found in the gigantic vault of stock imagery so I'd need somebody to help me with even that, but presuambly somebody with a recurring need for stock art probably doesn't lack those skills to the extent that I do.