How to scale intent, quality, and artistry with AI [video] | Hacker News Reader

How to scale intent, quality, and artistry with AI [video]

youtube.com

86 pointssimonjgreen33 comments

Many people have no interest at all in art, if it is not used for connection with others. Scale all you want, humanity overwhelming rejects it.

The end result is that certain genres of art (digital paintings, video games, digital music) will die, and people will return to participating in the real world. The purveyors of artistic expression will flow back towards those who perform in the real world, with different recombinations of the genre, tending towards those that can allow for recognition of genuine expression.

AI will kill certain genres of art, and many more genres of art will be born.

That's quite bleak, and I hold an opposing view. My main view is people are generally quite good to discern AI cruft from "human-made" paintings, games, and music. As soon as it will be impossible to make money from AI cruft, the AI cruft will disappear. As a (hobby) game developer myself, no AI tool I have used has ever made me feel "wow, this is great". It has always required multiple iterations of feedback. I tried tools like Meshy too. I ended up hiring a part-time 3D designer from Indonesia, who is a joy to work with.

We are good at distinguishing it, but that doesn’t mean the unmitigated flow of Super Intelligence-generated images, music, etc won’t degrade the spaces we share and exhaust people from filtering from the art that they do want.

I can tell an image is AI generated almost instantly, but it nonetheless stomped places like the Photographers of Threads channel. Not that it was great to begin with, but it was something. More and more I’m motivated to only discover photographers via photo books.

What if I made a machine that live streamed my thoughts (via brainwaves) as images/video, i.e. a window into the imagination? That would involve both AI (necessarily) and communication.

Delusional thinking, I don’t know a single person that plays video games in order to connect with the creator of the game. They definitely play them to connect with others who play the same game. The only people not playing games due to AI are people who make a political issue out of it and are virtue signaling to their group and a small number of people with a massive chip on their shoulder about automation in media.

If someone says AI art is good, I think the one saying that is AI too.

The problem isn't that you can't use AI to uplevel your creativity. It's that doing it slows you down, so you end up at a marked disadvantage, because others who just prompt "hey Claude, make me X" will produce 100 good enough slop-pieces in the time it takes you to finish one.

For the vast majority of paying customers or social media appreciators, that low-effort slop is sufficient. It's cheap, fast, and gets better with every passing month. So if your overriding goal is to scale to win in that marketplace, then you can only scale by giving up on quality and artistry. It's a race to the bottom.

I think the only way to win is to push back on the notion that you need to work on superhuman scale in the first place. You can differentiate yourself from AI by asserting you're human and that there's inherent value in human art. You'll find some customers willing to pay extra for that, just like some hire a woodworker instead of buying at Ikea. But it's going to be a tough market for artisans because fewer and fewer people are forced to buy the fruits of your intellectual labor.

In software engineering specifically, we gleefully decided that speed and cost are the only thing that matters and we celebrate "hey Claude, make me X" projects the same as we celebrate human ingenuity (case in point: HN front page). So for SWEs, even that "I'm a human" route may be unavailable.

I have to agree. This is especially noticeable in software; it’s impossible to distinguish manual work from the generated kind. There’s no point in appealing to the human element here, the buyer simply doesn’t see the difference. But fortunately, for now, we still have work.

Year after year, I keep seeing people talk about various races to the bottom.

And yet all the things (in particular the technology things) just keep getting better and better?

"Cheap-to-generate" mediocre content forces the floor expectation to rise, becoming the "minimum", which raises expectations of "current mediocre" further. Its largely invisible, slow process, where you can't gauge past "floor-level" expectation for content, because it implicitly assumed the 'current year' standard is the norm(while its evolution of expectations). In a few years, expectation for "mediocre slop" would likely include stuff that considered top-level today, but has some minor deficiencies and the "slop of years past" would be relegated to unusable/useless junk, like\ e.g. Stable Diffusion 1.5 images and GPT2 texts.

all the things keep getting better and better? Many people seem to feel all the things are getting worse. If your life is getting better, that's nice, but I would expect you're aware enough of the rest of humanity to know most people feel it isn't getting better for them?

Don’t see how scaling mediocre content is a winning strategy. Quality will go up and we all win.

That's a very one-dimensional view. Two things I'd encourage you to consider:

1. What is quality, exactly? An average person on Facebook prefers AI-generated images to real photos or human art. Is their preference a signal of quality, or is this actually optimizing for orthogonal metrics - agreeability, engagement, etc?

2. Do increases in quality always translate to "everyone winning"? For example, is it a net win if a marketing flyer for your local grocery store has 10% higher "quality" if it means that your next-door neighbor who used to design these is no longer employed? Even if you disregard the feelings of the neighbor, is it good for you? You stand to lose more than you're gaining (e.g., reduction in property values if they can no longer afford upkeep).

I don't want to paint AI as a net negative, but these dynamics are pretty complex. There's plenty of situations where AI is not upleveling quality, or where quality is comparatively unimportant.

Did you watch the video? Her point is to use AI's productivity gains to iterate faster on the creative aspects.

That's my point. When you use AI tools in a "responsible" way in your creative process, you lose the ability to (honestly) differentiate your work as AI-free, but you can't possibly win the competition with AI on speed or price unless you give up more control than you want. And in terms of customer preferences, most people are content with low-effort AI art.

As a SWE and also an art director;

firstly, great post and you're spot on. On the art side, it was already that way before AI art, just that the smallest of customers didn't know how to access it.

Coding is just as much like woodworking as art and design are. The core of the matter is intent, or what people call intentionality in the architecture.

What I tell clients is: You want your code to be consistent and maintanable, so someone fully understands the edge cases. Or you want the design to be consistent across the company. You could easily hire someone to do one art piece that was 90% on point, and then haphazardly hire someone else to do one that was 90% okay based on that... and very quickly the overall structure will decohere.

So, I'm not talking about someone selling socks on Etsy, but if you've already invested a few million in branding and/or code architecture, that is an asset you own and control. Do you really want to inject potential AI slop into that in a process that's going to degrade it further in each subsequent step, just to save a few bucks?

I also like how people are becoming more aware of the difference... this thread is enlightening[0]... but I believe that what a few people are conscious of, 90% of people are subconsciously aware of, and knowing how to manipulate subconscious associations with imagery has always been the true art and magic of graphic design. Design in particular is widely misunderstood, even by people who call themselves designers. It has almost nothing to do with making something look subjectively decent (the bare minimum), and everything to do with knowing more references than your audience knows they know. So when AI becomes a recognizable visual trope, we've won.

>> it's going to be a tough market for artisans because fewer and fewer people are forced to buy the fruits of your intellectual labor.

Once everyone has a WordPress site, who needs a web designer? It turns out a lot of people. And I've begun getting random queries from people who have vibe-coded apps asking me to "fix" them. When everyone has a bike, people dream of owning a car. When everyone has a car, people dream of owning a plane. What makes them tick isn't the velocity, it's the cachet.

[0] https://www.reddit.com/r/Portland/comments/1wwyqiw/the_made_...

> For the vast majority of paying customers or social media appreciators, that low-effort slop is sufficient. It's cheap, fast, and gets better with every passing month. So if your overriding goal is to scale to win in that marketplace, then you can only scale by giving up on quality and artistry. It's a race to the bottom

I've arrived at the conclusion, that we can probably do without software developers rather than graphic designers. The slopocalypse is oozing into the real-world, where more and more businesses advertise their services with AI-generated art, if we want to call it even art at all. It's either overloaded and nonsensical advertisements, horrendous and lovecraftian depictions of food, or corporate designs that just scream that everything the business does is just cheap and low-quality. My bet is on graphic designers having a small renaissance, because obviously AI-generated art will become a signal for cheap, lazy, and low-quality business practices.

"...Use the good judgment of the human and scale with the AI" this is the way.

The points made in this video were great and key to modern day development. Often times AI just copies the same formula that gives "good" results but those results now just scream AI such as purple gradients in hero sections. Being intentional with your design choices is a must which this video explains well.

She could have a great career in audiobook narration

Just stumbled on this on YouTube and was compelled to share. This is the best (relatively) short form piece I’ve seen on avoiding AI slop while simultaneously embracing the gifts of AI

instead, human slop with that ted talk format.

all for the following. 1. give a shit. 2. build a system and codify it 3. review iterate and recreate 4. be creative

With the caveat that I have not watched the video, the title itself rubs me the wrong way. Intent, quality, and artistry as I understand them are not things intended to "scale." You would not, for instance, as a musician, think "I made an album, yes. But how do I scale up to ten or a hundred albums?" You may well think "How can I scale my audience," but that is a different question entirely and not a matter of intent, quality, or artistry. Scaling art is, to me, a wrongheaded notion to begin with.

But what if… that’s changing? What if some next generation is so properly trained to find repeating any song after a first play to be so tedious that the new art form is just an endless effusive scaled-out stream of the artist’s brain?

Well, that's possible, but I think it's very unlikely.

Then again, if you had told me a few years ago that AI was right around the corner and that the first jobs it would take would be in graphic design, programming and advanced mathematics, I would have laughed in your face.

Honestly I think I disagree. If you take photography as an art for example, the "scale" was pretty technology constrained for a long while.

In what regard? How many pictures you can take, places you can be, how fast you can edit? Photographers all have an archive of thousands of photos. The great ones deliberate for months or years to produce a 60-picture photo book. I don’t see tech scaling that part of the process, because it’s fundamentally about their artistic vision, emotional impact, etc.

Maybe you could use tech to scale bottlenecks of commercial photographers doing advertising or wedding shoots, but we’re talking about art.

I’m not sure what “scaling” is even supposed to mean here. The video’s thumbnail and description advocate against “zombie UI” (whatever that means as well — probably the typical AI slop-style UI). The pressing issue is that there is a dearth of good (great) UI, scaling or not, and that was already the case in the years before the use of LLMs. We need more instances of really good UI before we can talk about scaling.

It's a good talk and worth watching. The title doesn't really reflect what the talk is about, which is more about how to not lose those things while using AI tools. The title could maybe even be described as a kind of trojan horse because the answer (to the question of how to scale), if you read between the lines, seems to be basically: don't. Slow down and be intentional about every detail, focus on the user, and don't lower your standards.

Her stuttering and unnatural pausing makes me feel like she's fighting existential dread while delivering this peppy "optimistic" brand strategy video.

[flagged]

"Please don't post shallow dismissals, especially of other people's work. A good critical comment teaches us something."

https://news.ycombinator.com/newsguidelines.html

I apologize. My comment was indeed shallow (ironically so).

Reply on news.ycombinator.com