The inevitable outcome here is that designed materials become so generic and infinitely produceable that they become worthless background noise. We are well on the way to that path.
For almost all materials the only value of getting a seriously produced work of design (i.e., the "make me a magazine-style pitch deck for our seed round" this design engine mentions) is a signaling function that some combination of effort and capital went into its production. Yes, the 1 in a 10,000 work of design adds some actual value. But usually it's just a filtering mechanism. The purpose of making a powerpoint deck before a meeting is rarely the value of a deck. Rather it is signaling that someone spent some time actually organizing their thoughts instead of bloviating spontaneously.
All of this is lost with AI led design. Producing designed artifacts are free and instant. Yeah you will impress the old folks for a year or so who haven't caught onto the joke. Eventually this just becomes pointless table stakes. Just the same way desktop publishing was in the 90s. You impressed the old folks for a bit until it all became background noise table stakes.
I’m glad people will have to evaluate the substance of the deck rather than using a cheap heuristic like how visually appealing the presentation is.
I understand there tends to be a correlation between visual appeal and effort, and correlation between effort and merit, but correlation is notoriously flawed. Flawed models can be useful, but only if one qualifies their use sufficiently. I don’t think most people who used are using the aesthetics heuristic you mention to gauge merit are using it rigorously to sharpen their thinking, they’re using it as a shortcut to prevent themselves from needing to think.
An equally plausible scenario to that of which you mention is that technical people can make presentations that are similarly visually appealing as the non-technical people, and that their opinions will be valued more than before. Maybe this will happen, maybe this won’t happen, but I am certain that we do not know yet.
> The purpose of making a powerpoint deck before a meeting is rarely the value of a deck. Rather it is signaling that someone spent some time actually organizing their thoughts instead of bloviating spontaneously.
> All of this is lost with AI led design. Producing designed artifacts are free and instant. Yeah you will impress the old folks for a year or so who haven't caught onto the joke.
I was at an AI/LLM themed hackathon recently. At the end the winning teams presented what they’d done.
The slides were all AI generated, which was fair given the theme and the short time they had at the end to prepare to say something (~10 minutes given to prepare after winners were announced, and before that all teams were spending all the three or so hours we had fully focused on the tasks rather than wasting time making presentations about what had been done).
Still felt a bit weird to see someone speak with slides that were as surprising to themselves as it was to the audience. Like I said, no shade on them in this case given the theme of the hackathon. But it does make me wonder how the future will be at many jobs where “velocity”/“productivity” is so much in the focus that unreviewed LLM generated slides becomes the norm. Hopefully not.
Obviously if you make the slides yourself then you'd know the content well.
The way of using these tools is not to one-shot your slide deck (unless you have plenty time to learn the content) but give it a base product you've already worked on and ask it to make it pretty, interesting, etc. and perhaps make small changes to the content which you'd review and learn.
You can probably use a knife as a fork but it wouldn't be the best way of using the knife.
I beg people to send me their prompts rather than the stochastic text expanded drivel they send me as memos/plans/etc... Massive waste of my time responding to ghosts - actually taking 10 pages seriously that often the "author" has barely read. I'd much rather get some unstructured bullet points if those are actually a person's ideas.
I love AI. Used well it's a massive enhancer to make things. But yeah whats the value of a presentation that the presenter is also seeing for the first time. Not just zero. Since it wasted everyone's time and bandwidth the value is negative.
I think design as a "signaling function" for determining the quality of a thing was already broken. It was already possible to put up an impressive-looking site for anything; already possible to to dupe people with cheap product wrapped in fancy packaging.
Movies with insane budgets that spend forever in production are often still terrible. One of my favorite songs was written by the artist in a hotel room on a Sunday afternoon.
One thing to consider: if it's cheap and immediate to wrap any content in design, it can now also be cheap and immediate to customize the design of content. Maybe we can finally return to a user-focused internet like the one that was promised to us by browser custom style sheets.
Finally, I can see democratizing design in this way will make more content more pleasant to look a (which is a win). And we'll also make better decisions with design out of the decision matrixes it doesn't belong in (another win).
One use of design is signaling, but not all - successful design is that which fulfills its purpose.
Many designed things do not need to be differentiated and will benefit from a homogenous AI-powered design (internal documentation, local service business communications, etc) in the same way that desktop publishing replaced hand or type-written notes but did not replace professional designers (although it did require them to learn digital tooling).
For designs that do benefit from being differentiated it'll be interesting to see what happens. If anything, AI homogeneity provides more opportunity for talented human designers who can provide "design alpha" beyond whatever trends the LLMs sucked up in their last crawl.
>The modern world is design rich and art poor. That is to say that with the introduction of mass production it became possible to distribute everywhere items that in previous eras would have been seen as full of Mana, but now, not unique — they have none.
That was in reference to mass produced design, but it seems to apply 10X to AI produced Art.
> The inevitable outcome here is that designed materials become so generic and infinitely produceable that they become worthless background noise.
Hopefully. The process has taken way too long. Compare to something similar like PowerPoint animations. Fun the first time you see them, and then annoying after that.
The best possible side effect of the cost of producing content dropping to zero would be more effort spent honing a message into its most concentrated form.
This is why I beg people to send me their raw prompts rather than the output of running that prompt through a generic LLM. Same semantic content in a half page prompt as the 10 page stochastic text expansion LLM memo version but 1. takes me 1/10th the time to process and 2. I'm not forced to guess which ideas are real ideas I need to respond to and which are just text expansion ghosts.
That idea around LLMs reducing the signaling value of certain types of work is very interesting, and I hadn’t really thought about it.
I think about this effect with targeted advertising a lot, every since I read this article about why targeted advertising is so useless for both consumers and advertisers, even when it seems like on the surface it should be better for both: https://zgp.org/targeted-advertising-considered-harmful/
Once it becomes very cheap to do something, it becomes completely useless as a way to differentiate quality from crap.
This is true if you that assume the only purpose of design is aesthetic differentiation. There actually is a lot of science in how you scan information in a design, how it's presented, the visual hierarchy, grouping and things that actually have utility in and off itself.
And the result will be people creating their own flawed but unique designs as a counter signal. Think of the early internet and the janky but wonderfully personal websites it spawned.
> That's not "AI tries to design something". That's an AI that has been trained, by the prompt stack, to behave like a senior designer with a working filesystem, a deterministic palette library, and a checklist culture
What, you don't want your senior designer to have a working filesystem and checklist culture? No deterministic palettes?
This insufferable period of AI will eventually come to an end. We just have to power through as people get fed by it. Mindless writeups, social media posts, emails, cold calls.. they all bear common trait of no to low effort and it shows. It's as vapid and empty as the impulse that thought it would be a good idea. Some of us consciously go by the rule that if you haven't bothered to write, we're not going to bother to read.. rest will grow into the same mindset. Spam farms pretending to be sales optimizations, linkedin lunatics with "valuable messages", low effort slackers larping to be engineers.. it's all going to go away and true value will prevail. Right now, it's not too much different than nigerian prince letters, industrialized.
To be fair I find the approach from claude design incredibly wasteful of tokens, and time-consuming since it needs to build a full website. Their website is also clearly vibe-coded and not homogeneous in style with the rest.
ChatGPT image 2 is much better at protoyping uis, cheaper and faster. I haven't tried the figma plugin but I suspect it's also more efficient.
Readme reads like a sales deck. Got to "six load-bearing ideas" and closed the tab. If your tool was actually good you'd just show what it does.
Also 14k stars in a week is doing a lot of work here. Nobody finds a repo that fast organically.
I'm curious what flows folks find most productive here? We are a heavy vibe coding team, with heavy review. That has smoothed out for our backend work, but frontend feels much earlier.
We have AI driving a usual mix of storybook, pencil, figma, playwright, tailwind/react, per-pr staging servers, etc, and a few skill files on using these. PRs include autogenerated storybook and intool screenshots, and links to staging servers.
Except... Everyone works quite differently in how they flow through this. Likewise, it's unclear how valuable each pieces still is, and when. Our developers are doing more ownership now, which is shifting this too.
Are folks switching to Claude Design? Some super skills imports? Etc..
We have a service issue today with the rate of tooling, it reminds me of the old days using Napster or Kazaa, it's full of good stuff but the curve to get in is so high, so much cognitive energy required to understand and get value out of it and, once you do, you have this monstrous workflow that only you know how to operate.
Or maybe I'm getting old? Serious question, do people really open a project with a README like that and are able to hit the ground running quickly and getting value right out of the gate?
Seems like this is an artifact / deliberate smoothing: Other tracker sites have unrounded data, and the github API does show a consistent growth instead of bursts at rounded intervals.
Once stars started becoming a marketable metric, it's pretty much inevitable that they would be purchasable. Same as any other bot review market I suppose
It depends on what you’re doing. If you’re working on a product like Slack, for example, the right question will often be not what UI your feature needs but what feature your UX idea needs.
curious how much of the output quality is the design systems and skill files doing real work vs claude just being very good at HTML. the prompt stack matters, but it's hard to know how much.
LLM-created designs are already recognizable and are the new Microsoft keynote templates. Boring, vapid, devoid of personality, perfectly fine for business use.
So as a design engine, sure. What things like this are trying to claim is that you can get "good" design and well, that's subjective. Y'know how people who don't understand kerning can look at bad kerning and feel something's not right but lack the words to explain why? The same goes for LLM design.
I'm not a luddite, I enjoy using Claude to assist in coding tasks but visual design will never be something I choose to use any LLM for. Design is for humans and LLMs lack taste.
How does a human designer even compete? I just looked at all the demos and they look beautiful.
I hand designed my site https://www.nair.sh/ and it feels like it doesn't even compare.
Sure, there's some judgment as to what design is appropriate in a given situation, but it just feels like so much harder for a human's design to feel valuable now.
I feel the designs they present are actually quite bad. Like... they are an anti-ad for this product. Just random fonts, bold, italics, underlines. Bad contrast, skinny small fonts.
Your site is actually really nice except the red color burns into my retina, so that's the only thing I would change about it (change your --primary to something more like #7c2c3e)
Originality. The same as with art. Art and design are more than just a mean to satisfy a need. They are an opportunity to explore, to question. When Georges Seurat developed pointillism, he wasn't trying to compete with the people who could imitate Raphael. He created his own direction.
The inevitable outcome here is that designed materials become so generic and infinitely produceable that they become worthless background noise. We are well on the way to that path.
For almost all materials the only value of getting a seriously produced work of design (i.e., the "make me a magazine-style pitch deck for our seed round" this design engine mentions) is a signaling function that some combination of effort and capital went into its production. Yes, the 1 in a 10,000 work of design adds some actual value. But usually it's just a filtering mechanism. The purpose of making a powerpoint deck before a meeting is rarely the value of a deck. Rather it is signaling that someone spent some time actually organizing their thoughts instead of bloviating spontaneously.
All of this is lost with AI led design. Producing designed artifacts are free and instant. Yeah you will impress the old folks for a year or so who haven't caught onto the joke. Eventually this just becomes pointless table stakes. Just the same way desktop publishing was in the 90s. You impressed the old folks for a bit until it all became background noise table stakes.
I’m glad people will have to evaluate the substance of the deck rather than using a cheap heuristic like how visually appealing the presentation is.
I understand there tends to be a correlation between visual appeal and effort, and correlation between effort and merit, but correlation is notoriously flawed. Flawed models can be useful, but only if one qualifies their use sufficiently. I don’t think most people who used are using the aesthetics heuristic you mention to gauge merit are using it rigorously to sharpen their thinking, they’re using it as a shortcut to prevent themselves from needing to think.
An equally plausible scenario to that of which you mention is that technical people can make presentations that are similarly visually appealing as the non-technical people, and that their opinions will be valued more than before. Maybe this will happen, maybe this won’t happen, but I am certain that we do not know yet.
I really love this take. AI both increases and decimates the ability of people to BS you with fancy graphics and text.
1 reply →
> The purpose of making a powerpoint deck before a meeting is rarely the value of a deck. Rather it is signaling that someone spent some time actually organizing their thoughts instead of bloviating spontaneously.
> All of this is lost with AI led design. Producing designed artifacts are free and instant. Yeah you will impress the old folks for a year or so who haven't caught onto the joke.
I was at an AI/LLM themed hackathon recently. At the end the winning teams presented what they’d done.
The slides were all AI generated, which was fair given the theme and the short time they had at the end to prepare to say something (~10 minutes given to prepare after winners were announced, and before that all teams were spending all the three or so hours we had fully focused on the tasks rather than wasting time making presentations about what had been done).
Still felt a bit weird to see someone speak with slides that were as surprising to themselves as it was to the audience. Like I said, no shade on them in this case given the theme of the hackathon. But it does make me wonder how the future will be at many jobs where “velocity”/“productivity” is so much in the focus that unreviewed LLM generated slides becomes the norm. Hopefully not.
Obviously if you make the slides yourself then you'd know the content well.
The way of using these tools is not to one-shot your slide deck (unless you have plenty time to learn the content) but give it a base product you've already worked on and ask it to make it pretty, interesting, etc. and perhaps make small changes to the content which you'd review and learn.
You can probably use a knife as a fork but it wouldn't be the best way of using the knife.
3 replies →
I beg people to send me their prompts rather than the stochastic text expanded drivel they send me as memos/plans/etc... Massive waste of my time responding to ghosts - actually taking 10 pages seriously that often the "author" has barely read. I'd much rather get some unstructured bullet points if those are actually a person's ideas.
I love AI. Used well it's a massive enhancer to make things. But yeah whats the value of a presentation that the presenter is also seeing for the first time. Not just zero. Since it wasted everyone's time and bandwidth the value is negative.
1 reply →
Look I'm speaking here as a career designer:
I think design as a "signaling function" for determining the quality of a thing was already broken. It was already possible to put up an impressive-looking site for anything; already possible to to dupe people with cheap product wrapped in fancy packaging.
Movies with insane budgets that spend forever in production are often still terrible. One of my favorite songs was written by the artist in a hotel room on a Sunday afternoon.
One thing to consider: if it's cheap and immediate to wrap any content in design, it can now also be cheap and immediate to customize the design of content. Maybe we can finally return to a user-focused internet like the one that was promised to us by browser custom style sheets.
Finally, I can see democratizing design in this way will make more content more pleasant to look a (which is a win). And we'll also make better decisions with design out of the decision matrixes it doesn't belong in (another win).
It could also be that It just shifts the burden from execution to strategy.
It's not enough anymore to have someone push nice pixels for you.
You'll need to consider if your design aligns with how you want to position yourself wrt to other players in your space.
Human communication moves ever closer to its final form: bullet pointed lists of lower case text and emoji
One use of design is signaling, but not all - successful design is that which fulfills its purpose.
Many designed things do not need to be differentiated and will benefit from a homogenous AI-powered design (internal documentation, local service business communications, etc) in the same way that desktop publishing replaced hand or type-written notes but did not replace professional designers (although it did require them to learn digital tooling).
For designs that do benefit from being differentiated it'll be interesting to see what happens. If anything, AI homogeneity provides more opportunity for talented human designers who can provide "design alpha" beyond whatever trends the LLMs sucked up in their last crawl.
https://medium.com/luminasticity/art-as-a-tool-for-storing-m...
>The modern world is design rich and art poor. That is to say that with the introduction of mass production it became possible to distribute everywhere items that in previous eras would have been seen as full of Mana, but now, not unique — they have none.
That was in reference to mass produced design, but it seems to apply 10X to AI produced Art.
> The inevitable outcome here is that designed materials become so generic and infinitely produceable that they become worthless background noise.
Hopefully. The process has taken way too long. Compare to something similar like PowerPoint animations. Fun the first time you see them, and then annoying after that.
The best possible side effect of the cost of producing content dropping to zero would be more effort spent honing a message into its most concentrated form.
This is why I beg people to send me their raw prompts rather than the output of running that prompt through a generic LLM. Same semantic content in a half page prompt as the 10 page stochastic text expansion LLM memo version but 1. takes me 1/10th the time to process and 2. I'm not forced to guess which ideas are real ideas I need to respond to and which are just text expansion ghosts.
The medium becomes the message
That idea around LLMs reducing the signaling value of certain types of work is very interesting, and I hadn’t really thought about it.
I think about this effect with targeted advertising a lot, every since I read this article about why targeted advertising is so useless for both consumers and advertisers, even when it seems like on the surface it should be better for both: https://zgp.org/targeted-advertising-considered-harmful/
Once it becomes very cheap to do something, it becomes completely useless as a way to differentiate quality from crap.
This is true if you that assume the only purpose of design is aesthetic differentiation. There actually is a lot of science in how you scan information in a design, how it's presented, the visual hierarchy, grouping and things that actually have utility in and off itself.
Everything is already pretty homogenous I wouldn't blame AI for this.
Everything goes back to plain text. Would be kind of nice.
I don't disagree, but I'm not sure I see the point you're trying to make.
Maybe ask your LLM to explain it?
1 reply →
True. But this makes it easier to stand out in a sea of monotony.
It's cheap. That's all.
And the result will be people creating their own flawed but unique designs as a counter signal. Think of the early internet and the janky but wonderfully personal websites it spawned.
The README is unnerving. Do people really see the Claude-salesman style of writing as something normal?
On the other hand, I should be thanking Anthropic for making it so easy to spot, they might have done this intentionally.
> That's not "AI tries to design something". That's an AI that has been trained, by the prompt stack, to behave like a senior designer with a working filesystem, a deterministic palette library, and a checklist culture
What, you don't want your senior designer to have a working filesystem and checklist culture? No deterministic palettes?
What is a deterministic palette, and why is it so important here?
> OD stands on four open-source shoulders:
That's impressive, although I'd hesitate to call that "standing", it's more a crawl I'd say.
I hate when my palette library is probabilistic.
This insufferable period of AI will eventually come to an end. We just have to power through as people get fed by it. Mindless writeups, social media posts, emails, cold calls.. they all bear common trait of no to low effort and it shows. It's as vapid and empty as the impulse that thought it would be a good idea. Some of us consciously go by the rule that if you haven't bothered to write, we're not going to bother to read.. rest will grow into the same mindset. Spam farms pretending to be sales optimizations, linkedin lunatics with "valuable messages", low effort slackers larping to be engineers.. it's all going to go away and true value will prevail. Right now, it's not too much different than nigerian prince letters, industrialized.
Yep I agree. I was looking for a getting started like for example here for openspec: https://github.com/Fission-AI/OpenSpec/blob/main/docs/gettin... but couldn't find anything like that
I found the At a glance section especially funny. Just a ton of buzzwords compressed together. One of the most dense tables I've ever seen on GitHub.
Agreed, I could just about bear it until I hit the “ Six load-bearing ideas” section. Very off-putting.
No thing-you-don't-want. No second-thing-you-don't want. Just thing-you-want, the-way-you-want-it.
It's midness. People don't find it good, it's just less effort to meh it.
It's bad, not mid. And it's insulting to the reader.
>Do people really see the Claude-salesman style of writing as something normal?
In certain circles, yeah. It's bad powerpoint writing by ambitious but dull mid-level managers, memeified. There's a lot of it out there.
If you were to distill that kind of copy into an AI model and then reproduce it with just a touch of uncanny valley, yup. 100% that's what it is.
To be fair I find the approach from claude design incredibly wasteful of tokens, and time-consuming since it needs to build a full website. Their website is also clearly vibe-coded and not homogeneous in style with the rest.
ChatGPT image 2 is much better at protoyping uis, cheaper and faster. I haven't tried the figma plugin but I suspect it's also more efficient.
Readme reads like a sales deck. Got to "six load-bearing ideas" and closed the tab. If your tool was actually good you'd just show what it does. Also 14k stars in a week is doing a lot of work here. Nobody finds a repo that fast organically.
I'm curious what flows folks find most productive here? We are a heavy vibe coding team, with heavy review. That has smoothed out for our backend work, but frontend feels much earlier.
We have AI driving a usual mix of storybook, pencil, figma, playwright, tailwind/react, per-pr staging servers, etc, and a few skill files on using these. PRs include autogenerated storybook and intool screenshots, and links to staging servers.
Except... Everyone works quite differently in how they flow through this. Likewise, it's unclear how valuable each pieces still is, and when. Our developers are doing more ownership now, which is shifting this too.
Are folks switching to Claude Design? Some super skills imports? Etc..
We have a service issue today with the rate of tooling, it reminds me of the old days using Napster or Kazaa, it's full of good stuff but the curve to get in is so high, so much cognitive energy required to understand and get value out of it and, once you do, you have this monstrous workflow that only you know how to operate.
Or maybe I'm getting old? Serious question, do people really open a project with a README like that and are able to hit the ground running quickly and getting value right out of the gate?
Repo's been up for a week and already it has 14k stars.
Oh look, they are gaining stars at a rate of pretty much exactly 1400 per day: https://www.star-history.com/?repos=nexu-io%2Fopen-design&ty...
Yeah, nothing shady here at all.
Wow, rounded exactly to the hundreds? Or is that an artifact of star-history's data-collection?
Seems like this is an artifact / deliberate smoothing: Other tracker sites have unrounded data, and the github API does show a consistent growth instead of bursts at rounded intervals.
https://www.startrail.dev/?repos=nexu-io%2Fopen-design&legen...
The growth remains very artificial, a steady line instead of the real jitter than real stars would have.
Do people really buy stars for github? Would explain some of these crazy growths
Don't underestimate stars given by claws without direct instruction from humans. Not bought or sold, but not real either.
Once stars started becoming a marketable metric, it's pretty much inevitable that they would be purchasable. Same as any other bot review market I suppose
Apparently so. Posted 12 days ago: https://news.ycombinator.com/item?id=47831621
2 replies →
Do people design UIs first?
I just basically define what I need in a UI in plain text
when the prototype is built.
I extract the repeating units, then add design to it.
If you're product first, you design the UX, which includes the UI.
If you're tech first, you do what you do.
It depends on what you’re doing. If you’re working on a product like Slack, for example, the right question will often be not what UI your feature needs but what feature your UX idea needs.
This seems bloated to me
This seems like a lot and most people don’t need such as overly complex solution solving for both UI, UX and marketing.
I created something slightly simpler that just generates a token system and allocates the UX to the LLM.
github.com/bmson/anchor-ui
This feels so over complex. I did something similar but went with something quite a bit simple
https://github.com/bmson/anchor-ui
curious how much of the output quality is the design systems and skill files doing real work vs claude just being very good at HTML. the prompt stack matters, but it's hard to know how much.
LOL
LMAO, even.
LLM-created designs are already recognizable and are the new Microsoft keynote templates. Boring, vapid, devoid of personality, perfectly fine for business use.
So as a design engine, sure. What things like this are trying to claim is that you can get "good" design and well, that's subjective. Y'know how people who don't understand kerning can look at bad kerning and feel something's not right but lack the words to explain why? The same goes for LLM design.
I'm not a luddite, I enjoy using Claude to assist in coding tasks but visual design will never be something I choose to use any LLM for. Design is for humans and LLMs lack taste.
I don't think I've ever wanted a README to fuck off more than this one, impressive.
Linux version?
[dead]
[dead]
[dead]
nice work!
[dead]
[dead]
[flagged]
I really appreciate open source community for moving this fast
How does a human designer even compete? I just looked at all the demos and they look beautiful.
I hand designed my site https://www.nair.sh/ and it feels like it doesn't even compare.
Sure, there's some judgment as to what design is appropriate in a given situation, but it just feels like so much harder for a human's design to feel valuable now.
We are soon going to converge on all websites looking exactly the same, we’re almost there really
It’s just the same sterile template used for everything, yeah it looks good first time you See it. But the 100th? It starts to look like noise
Are you a designer? Everything AI does looks impressive if you are not familiar with it.
You're right in that our expertise can see how this was not generated with the same kind of thoughtfulness that we might apply.
But you're wrong in implying (if you are) that it's not valuable to be impressive to a non-expert.
5 replies →
I feel the designs they present are actually quite bad. Like... they are an anti-ad for this product. Just random fonts, bold, italics, underlines. Bad contrast, skinny small fonts.
Your site is actually really nice except the red color burns into my retina, so that's the only thing I would change about it (change your --primary to something more like #7c2c3e)
How do human artists compete with AI-gen images?
Yes, we're building a dystopia where AIs do the work humans enjoy, and humans get to hold on to drudgery. What's not to like?
2 replies →
Your point? It's an analogical problem.
I love writing but even there I have to work doubly hard to make sure I'm doing something valuable.
My point is that the space within which human creators can distinguish themselves is diminishing rapidly.
Originality. The same as with art. Art and design are more than just a mean to satisfy a need. They are an opportunity to explore, to question. When Georges Seurat developed pointillism, he wasn't trying to compete with the people who could imitate Raphael. He created his own direction.
Yes but you're talking about groundbreaking work.
There's so much joy to be found in regular human creating and sharing.
The creating part still remains because it's intrinsic but the sharing part feels discouraging now.
Regular, non-groundbreaking creative work seems ... less worthy of sharing?
5 replies →