8 ms·

AI and the Rise of Mediocrity

66 points by CharlesW 3y ago

andsoitis 3y ago
> The real danger to human creativity that these tools represent is the mechanization of human innovation. Relying on these tools will discourage us from looking beyond what has been done before, and further reduce innovation into no more than imitative remixing. Businesses and corporations assert that mechanization will benefit both the worker and the consumer. It’ll save time and make things more efficient. Efficiency holds back innovation and creativity. It focuses on cost at the expense of discovery and exploration.
- aik 3y ago
  That quote: I don’t buy it. There is plenty of incremental innovation now and more ambitious people still shoot for the stars. If AI is mediocre and doesn’t solve what humans want to solve, it won’t replace human desire for real innovation. Ideas are cheap and plentiful, everyone has them. They will be pursued, just in greater numbers.
  - 23B1 3y ago
    Most of the incremental innovation I've seen from tech for the past ten years has been darkpatterns, enshittification, surveillance, and profit-driven exploitation.
- founderspend 3y ago
  No it doesn't. Efficiency unlocks bandwidth for us to explore on higher levels and new directions, and it always has. Efficiency of energy production, manufacturing, transportation, information sharing etc – you cannot seriously believe that these did not spark new waves of innovation and creativity.
jruohonen 3y ago
"The problem for AI is that creative work is not predictable. It is not about statistical likelihood or simply mashing up the familiar—it is about leaps in logic and counterintuitive juxtapositions. It is about the unique experience of the individual, and seeking to do what has never been done before. It is about the least predictable next word or pixel. So the danger is not that AI programs will write the next great novel or create the next great painting, successfully replacing human inventiveness: they never will. The greater danger is that they won’t need to create great writing or art."
- yankoff 3y ago
  I never get this argument. There is no magic in the human creative process. One way or another, algorithms should be able to replicate it and eventually surpass it.
  - attilakun 3y ago
    They already did in some domains, for example AlphaGo's move 37. The article's author is coping hard.
  - morelisp 3y ago
    The article and excerpt are about the AI systems that do actually exist, not the ones that might someday exist.
    - yankoff 3y ago
      Author explicitly says: "they never will"
      - morelisp 3y ago
        "They" refers to the programs that exist (or more broadly anything based on "statistical likelihood or mashing up the familiar", which is the overwhelmingly dominant sort); "never will" refers to "never will write the next great novel or create the next great painting", not "never will exist." I think the author might agree that they never will exist, but because of his actual point about what they will do to creative work and communities, not any physical limitations.
  - jruohonen 3y ago
    > I never get this argument. I think it is still a solid argument, especially when juxtaposed with anything related to developing something new and innovative, whether in the domain of arts, science, or business. But what I found insightful in this quote was the note about the least predictable things.
  - 3y ago
noufalibrahim 3y ago
There was a book by Roald Dahl called the great automatic grammatizer which described an automatic typewriter that could be tuned to automatically create novels. Mediocre but quickly producible content swamped the market and genuine authors found it hard to get paid and so turned to these automatons to ghost write for them. A few held out and the cost was starvation. The story ends with "give us the courage to watch our children starve". It was prescient. And i think this article captures the same sentiment.
- simbolit 3y ago
  > mediocre but quickly producible content swamped the market I hear stuff like this all the time. It feels odd, given that pulp fiction dimeback "novels" exist for over a century. Some of these series run into several thousand issues, with authors expected to write a new novel each week. The "mediocre but quickly producible content" has existed for ever, and it hasn't drowned out the "serious" novel market. Why should it now?
  - mdgrech23 3y ago
    I'm not sure if you're American but take a look at the food situation here. Eating out practically anywhere is total garbage food. If you want high quality stuff made by people who know what they're doing be prepared to pay big bucks. A little pastry shop opened by me with a professionally trained French chef. The croissants are out this work but are like 8 bucks a piece whereas the Costco ones are maybe a dollar a piece and taste good. The market obviously incentivizes consumers to purchase the Costco variety but that's b/c the Costco variety is being more heavily subsidized than people realize.
    - fallingknife 3y ago
      Your $8 croissants are subsided just as much as the $1 ones at Costco. They are cheaper because they don't have to pay that professionally trained french chef.
      - 3y ago
spacecadet 3y ago
Time has run out... "AI", code gen, etc. have only amplified my creativity and output, soooo if you are feeling mediocracy creeping in, it always was. Sorry. If anything this has only highlighted something I have said for 40 years. Ideas are a dime a dozen, everyone is mediocre(including me), execute. Now, we will see who has the real ability to produce...
- JofArnold 3y ago
  Mediocre ideas are a dime a dozen. Meanwhile to me at least it seems as if good ideas are increasingly rare.
  - AlexandrB 3y ago
    My idea: Uber for ideas.
    - spacecadet 3y ago
      AWS for ideas, scale the ideas!!
- dorkwood 3y ago
  I agree with you 100%. Recently, I've been generating landscapes with Midjourney, signing my name at the bottom, and hanging them on my wall. It feels good to finally overcome the artistic challenges I faced in the past.
  - gumballindie 3y ago
    But you are still not an artist, lack skill and the "art" has zero value.
    - peebeebee 3y ago
      He might have never been, but he's more of an artist now than he was before.
mlboss 3y ago
I don’t buy it. Humans are also a prediction machines. We continuously predict next thoughts, feelings, physical motion etc. I think what llms lack right now is: 1. Continuous input from through different senses 2. Ability to do experiments 3. More modalities like touch, sound etc Once these capabilities are added then they will be comparable to human “creativity”
- ggpsv 3y ago
  That's quite the jump in #2. What guides the experiments?
freemanon 3y ago
It's maybe hard to fathom for individualists, but it's probably humans that are quite mediocre on average. These AI models simply represent that. And on the value of generative AI, the same argument against new trends has been made on previous inventions. For decades, a typewriter was still more practical than using a word processor on a computer. For more than a decade, calls and SMS were more practical than online messengers and forums.
shmatt 3y ago
Top 10 summer movies 1982-1989 - 25% are sequels. In the 2010's that ticked to over 50%[1] This is way beyond AI. Culture just repeats itself, from movies to fashion to music. In case anyone was wondering - in the never ending loop between oversized baggy jeans and skinny jeans, we are right at the very end of the baggy era, looping back to skinny in a year or 2 So is this article complaining that instead of looping between baggy and skinny we should all be wearing silver spandex space suits? I guess thats true, but also Im good with just jeans personally [1] https://www.nytimes.com/interactive/2019/05/24/business/media/summer-sequels.html https://www.nytimes.com/interactive/2019/05/24/business/medi...
- gumballindie 3y ago
  > Culture just repeats itself, from movies to fashion to music. And including AI. If you watch movies, from around 90-95 you will notice the same talk about AI taking over the world, jobs, and all the schizofrenia surroinding it.
- 23B1 3y ago
  You've confused a loop with a spiral. Yes, culture is often derivative and yes sequels happen – but also culture builds and shifts and changes unpredictably and delightfully.
DonnyV 3y ago
That article was interesting but holy crap what was the deal with the constant shifting of the layout. Ads kept popping in and out shifting the content up and down. Its unreadable.
23B1 3y ago
I'm not worried in the medium-to-long term. Tech (which is nothing more than a lackey of Big Business) will never understand that real demand for cultural artifacts – art, books, film – comes from the consumers' need for real human connection. (The rare exception might be a guy like Steve Jobs). This is a huge advantage to artists and creative people because they can and will bypass that mindset entirely, and can do so with the confidence that they're connected to what human beings – consumers – actually want. It's why 'tech forward' orgs like NFLX can't really do much except churn out derivative works, meanwhile you've got the Davids, the Wachowskis, the Cohens, Tarantino, Chris Nolan, etc. constantly pushing the limits of creativity and writing their own ticket, without the permission of big business and its lackey tech bros. When these guys 'break out' of course everyone wants to be their best friend, but in no way was Warner Brothers or Paramount able to mechanize or catalyze that creative genius. What chaps my ass the most about the current crop of GAI is that its gone after writers and illustrators and artists because, in the process of sharing their works with the world, had that work hoovered up and used for training data without their permission or recompense. Luckily, all of the mediocrity these systems shit out is obviously superficial and soulless. Yeah you might some day be able to do a fully reactive murder mystery in Zuck's metaverse, but eventually you're going to want to be intrigued and delighted by something truly fresh and human – and that can only be created by other humans.
mjr00 3y ago
> A few days later I found myself in a grocery checkout line, skimming through yet another article in which the writer touted the wonders of “artificial intelligence” and fretted hazily over whether we are nearing the point when AI will be able to produce novels, films, and other creative work, effectively replacing us. When I looked up and over to other people in the line, half of them wore the same shoe brand as me. This is an interesting phenomenon with AI. People immediately assume that because AI has the ability to flood the market with mediocre art/writing/whatever, that artists and writers are screwed. In my experience it's quite the opposite; when the market gets flooded with repetitive mediocre garbage, the relative scarcity of something unique increases and actually increases the price, accordingly. To the author's point, you can get the same Adidas shoe as everyone else for $50, or you can get a largely similar-looking pair of Guccis for ~$800 US[0]. Those Guccis would not exist if not for the mass-market cheap option; there would be no reason to splurge on an expensive unique purchase if everyone else was already wearing something unique. Or not wearing shoes at all. This has been my experience since the introduction of stable diffusion and dall-e as well. Just anecdotal, as I don't have any access to hard data, but the artists I've spoken to say that their commissions and fanbox subscriptions have gone up since SD was released. Much like the trope of friend groups keeping an ugly friend around to make themselves look better, the mediocre and soulless AI art that's flooding the internet has, apparently, made people appreciate human output as a premium brand. [0] https://www.nordstrom.com/s/gucci-new-ace-sneaker-women/4701092 https://www.nordstrom.com/s/gucci-new-ace-sneaker-women/4701...
- Workaccount2 3y ago
  AI art isn't as good as human art. But in a few years it will probably be better. It's a foolish mistake to look at the state of AI today and say "Actually this is gonna be great for my creative business!" What is scary about AI is not ChatGPT or Midjourney, Dalle, whatever. What is scary is that they showed up 10 years sooner than the earliest predictions, and are still rapidly improving. That is what should give people pause and it's amazing how many people miss this.
  - mjr00 3y ago
    > AI art isn't as good as human art. But in a few years it will probably be better. This is a very big "probably". About as big as the "in a few years, AI drivers will probably be better than human drivers" line from 2015. Regardless: an AI can definitively be better than a human at something like Go (a closed game with a clear win condition) or driving (a logistical task with no creativity). But how do you define "better" when it comes to art? From a technical perspective, AI is already better than the vast majority of human artists! Very few artists would be able to, say, draw images of nature to create something that has the same contours as a popular meme[0]. So, arguably, we're already at the point where AI-generated art is "better" than human art. The question is, does that matter? Or does it matter in the same way that it matters that a motorcycle is faster than Usain Bolt? [0] https://i.imgur.com/jsI0HZH.png https://i.imgur.com/jsI0HZH.png
Fripplebubby 3y ago
LLMs will predict the most likely output to the prompt input - does this mean that a hypothetical perfect LLM would produce the most generic and widely-acceptable response to any input, the "lowest common denominator" output? The author seems to think so, and indeed, with today's GPTs, that is often what you might get back. But I think there's a lot more going on here - with natural language, a lot of the training effort goes into producing natural language output which conforms to the formal rules of language (it is intelligible, syntactically correct or roughly so, follows grammar, and semantically "about" what it is supposed to be about). But the interesting thing about natural language is that there are infinitely many correct responses which are valid formal responses to the prompt, both ones that introduce new or rare creative possibilities and ones that tread the same ground as a wikipedia article, for example (which is usually intended to be very neutral). So an LLM can produce both types of responses - formally correct and boring, formally correct and interesting - and you can see this today by offering "mix-ins" to the prompt like, "Tell me about Saudi Arabia's ruling family, in the style of Peter Griffin". Of course, it won't select to use the style of Peter Griffin without that prompt, but I think you can imagine a "weirdo" LLM that is trained with some positive response toward creative or interesting output too.
- Fripplebubby 3y ago
  I don't mean to imply that creativity is something that can be quantified objectively (used as the objective function, even), but it is widely understood that LLMs can produce varied output and be given particular biases or personalities
- hansvm 3y ago
  LLMs don't generally predict the most likely output, except when people go in to tweak temperature and other settings to make them more predictable for certain more mechanical tasks. The whole point of an LLM is that its output is calibrated; it produces a text with the same conditional probability you would expect seeing that text in the wild (give or take that we still don't understand why neural networks generalize).
  - Fripplebubby 3y ago
Workaccount2 3y ago
Are there gonna be enough hats for all these people to eat in a few years? I think the perpetual image in media that humans are a unique entity with a magical inner force has gotten to everyone's head. It's just a fiction guys. It strokes egos to sell copies. We are meat computers outputting computation. There is nothing special inside. On our current trajectory in a few years AI will be far more creative than any human. Then we'll have to trudge through grumblings about how it feels so "cold and inhuman" despite blind studies showing people greatly prefer AI works.
- markwkw 3y ago
  I think there is something special. Some special coupling of emotional states between us (manifests as empathy, feeling of community, feeling of connection). Not saying that this connection cannot be replicated or faked, this remains an open question. Perhaps building an AI vastly smarter than us will conflict with replicating this coupling. Perhaps we will prefer artwork from an AI with which we do feel this type of connection. Perhaps this will make us build an AI that is really close to us in terms of experiencing feelings an emotional states, even if it means it's less "smart". This wouldn't be a dull outcome...
lhnz 3y ago
Hm, my own intuition is that it'll raise the baseline experience but then cause hyper-competition by human creators to exploit its powers to the limits while infusing further human creativity on top of what is mechanistically produced. I do not think this will result in overall mediocrity. It'll lift those with no access to art or way to pay for human creativity away from absolute poverty, while allowing those that already had access to these things a more competitive market.
Raidion 3y ago
I've posted about this before, but this is just the hollowing out of the "middle" of the value curve. Most people want something that's good enough, and they're not willing to pay for something that's N+1 good at 2N price. Other people want something that's N+5 good and will pay 3N price. As AI pushes down the lower bound of cost for creativity, the "mediocre" content creators/programmers/writers are going to get squeezed. At the same time, this means there is an untapped market for the people who can truly be creative and do great work. Thinking the difference between the Sports Illustrated AI articles and premium sports journalism (The Athletic), blog spam recipes vs America's Test Kitchen, or CNN/Fox front page vs NYT/WaPo/Politico, etc. You're going to have to vote with your dollar to support the things you care about. If you're not willing to pay for the higher quality, you're going to get AI (or other "budget") content that's making it's money by just being good enough to get you to see the advertisements. It's a free market out there and you need to be aware that just a few percent of people (probably roughly similar to the lurker/poster ratios) are the ones that enable high quality content to be successful. The HackerNews crowd are significantly more likely to be able to actively support creative pursuits that they find valuable, so I'd suggest you do so! Otherwise, when we don't reward quality, why wouldn't people just be mediocre?
bbotond 3y ago
It's ironic how the blandest, most predictable opinion piece is about how AI produces medicore content.
- lcuff 3y ago
  Totally agree. Insipid. No real insights, nuances or ideas.
thefz 3y ago
Rise? With the advent of social media, we have already witnessed a sea of mediocrity flooding the internet.
garethsprice 3y ago
Mediocrity is a good thing in the context of commodity goods. To use the author's example, you CAN still get unique custom made shoes and still buy hand-grown heirloom tomatoes at the farmer's market. The shoes will cost you $1,000+ and you'll pay a couple of bucks a tomato - prices which adjusted for inflation haven't changed since such practice was the only way people got shoes and tomatoes, and they got a lot less of them. Mediocrity and mass-production have made shoes and tomatoes accessible to everyone at far lower cost, and are not inherently bad.
tim333 3y ago
AI may have the opposite effect to the "five cars identical in make, model, year, and color", "same shoe brand" stuff in that you'll be able to have it design something unique for you. Already you can do that with digital artwork and it may extend to customising physical objects soon.

66 points by CharlesW 3y ago

> The real danger to human creativity that these tools represent is the mechanization of human innovation. Relying on these tools will discourage us from looking beyond what has been done before, and further reduce innovation into no more than imitative remixing. Businesses and corporations assert that mechanization will benefit both the worker and the consumer. It’ll save time and make things more efficient. Efficiency holds back innovation and creativity. It focuses on cost at the expense of discovery and exploration.

aik 3y ago
That quote: I don’t buy it. There is plenty of incremental innovation now and more ambitious people still shoot for the stars. If AI is mediocre and doesn’t solve what humans want to solve, it won’t replace human desire for real innovation. Ideas are cheap and plentiful, everyone has them. They will be pursued, just in greater numbers.
- 23B1 3y ago
  Most of the incremental innovation I've seen from tech for the past ten years has been darkpatterns, enshittification, surveillance, and profit-driven exploitation.
founderspend 3y ago
No it doesn't. Efficiency unlocks bandwidth for us to explore on higher levels and new directions, and it always has. Efficiency of energy production, manufacturing, transportation, information sharing etc – you cannot seriously believe that these did not spark new waves of innovation and creativity.

jruohonen 3y ago

"The problem for AI is that creative work is not predictable. It is not about statistical likelihood or simply mashing up the familiar—it is about leaps in logic and counterintuitive juxtapositions. It is about the unique experience of the individual, and seeking to do what has never been done before. It is about the least predictable next word or pixel. So the danger is not that AI programs will write the next great novel or create the next great painting, successfully replacing human inventiveness: they never will. The greater danger is that they won’t need to create great writing or art."

yankoff 3y ago
I never get this argument. There is no magic in the human creative process. One way or another, algorithms should be able to replicate it and eventually surpass it.
- attilakun 3y ago
  They already did in some domains, for example AlphaGo's move 37. The article's author is coping hard.
- morelisp 3y ago
  The article and excerpt are about the AI systems that do actually exist, not the ones that might someday exist.
  - yankoff 3y ago
    Author explicitly says: "they never will"
    - morelisp 3y ago
      "They" refers to the programs that exist (or more broadly anything based on "statistical likelihood or mashing up the familiar", which is the overwhelmingly dominant sort); "never will" refers to "never will write the next great novel or create the next great painting", not "never will exist." I think the author might agree that they never will exist, but because of his actual point about what they will do to creative work and communities, not any physical limitations.
- jruohonen 3y ago
  > I never get this argument. I think it is still a solid argument, especially when juxtaposed with anything related to developing something new and innovative, whether in the domain of arts, science, or business. But what I found insightful in this quote was the note about the least predictable things.
- 3y ago

noufalibrahim 3y ago

There was a book by Roald Dahl called the great automatic grammatizer which described an automatic typewriter that could be tuned to automatically create novels. Mediocre but quickly producible content swamped the market and genuine authors found it hard to get paid and so turned to these automatons to ghost write for them. A few held out and the cost was starvation. The story ends with "give us the courage to watch our children starve". It was prescient. And i think this article captures the same sentiment.

simbolit 3y ago
> mediocre but quickly producible content swamped the market I hear stuff like this all the time. It feels odd, given that pulp fiction dimeback "novels" exist for over a century. Some of these series run into several thousand issues, with authors expected to write a new novel each week. The "mediocre but quickly producible content" has existed for ever, and it hasn't drowned out the "serious" novel market. Why should it now?
- mdgrech23 3y ago
  I'm not sure if you're American but take a look at the food situation here. Eating out practically anywhere is total garbage food. If you want high quality stuff made by people who know what they're doing be prepared to pay big bucks. A little pastry shop opened by me with a professionally trained French chef. The croissants are out this work but are like 8 bucks a piece whereas the Costco ones are maybe a dollar a piece and taste good. The market obviously incentivizes consumers to purchase the Costco variety but that's b/c the Costco variety is being more heavily subsidized than people realize.
  - fallingknife 3y ago
    Your $8 croissants are subsided just as much as the $1 ones at Costco. They are cheaper because they don't have to pay that professionally trained french chef.
    - 3y ago

spacecadet 3y ago

Time has run out... "AI", code gen, etc. have only amplified my creativity and output, soooo if you are feeling mediocracy creeping in, it always was. Sorry. If anything this has only highlighted something I have said for 40 years. Ideas are a dime a dozen, everyone is mediocre(including me), execute. Now, we will see who has the real ability to produce...

JofArnold 3y ago
Mediocre ideas are a dime a dozen. Meanwhile to me at least it seems as if good ideas are increasingly rare.
- AlexandrB 3y ago
  My idea: Uber for ideas.
  - spacecadet 3y ago
    AWS for ideas, scale the ideas!!
dorkwood 3y ago
I agree with you 100%. Recently, I've been generating landscapes with Midjourney, signing my name at the bottom, and hanging them on my wall. It feels good to finally overcome the artistic challenges I faced in the past.
- gumballindie 3y ago
  But you are still not an artist, lack skill and the "art" has zero value.
  - peebeebee 3y ago
    He might have never been, but he's more of an artist now than he was before.

mlboss 3y ago

I don’t buy it. Humans are also a prediction machines. We continuously predict next thoughts, feelings, physical motion etc. I think what llms lack right now is: 1. Continuous input from through different senses 2. Ability to do experiments 3. More modalities like touch, sound etc Once these capabilities are added then they will be comparable to human “creativity”

ggpsv 3y ago
That's quite the jump in #2. What guides the experiments?

freemanon 3y ago

It's maybe hard to fathom for individualists, but it's probably humans that are quite mediocre on average. These AI models simply represent that. And on the value of generative AI, the same argument against new trends has been made on previous inventions. For decades, a typewriter was still more practical than using a word processor on a computer. For more than a decade, calls and SMS were more practical than online messengers and forums.

shmatt 3y ago

Top 10 summer movies 1982-1989 - 25% are sequels. In the 2010's that ticked to over 50%[1] This is way beyond AI. Culture just repeats itself, from movies to fashion to music. In case anyone was wondering - in the never ending loop between oversized baggy jeans and skinny jeans, we are right at the very end of the baggy era, looping back to skinny in a year or 2 So is this article complaining that instead of looping between baggy and skinny we should all be wearing silver spandex space suits? I guess thats true, but also Im good with just jeans personally [1] https://www.nytimes.com/interactive/2019/05/24/business/media/summer-sequels.html https://www.nytimes.com/interactive/2019/05/24/business/medi...

gumballindie 3y ago
> Culture just repeats itself, from movies to fashion to music. And including AI. If you watch movies, from around 90-95 you will notice the same talk about AI taking over the world, jobs, and all the schizofrenia surroinding it.
23B1 3y ago
You've confused a loop with a spiral. Yes, culture is often derivative and yes sequels happen – but also culture builds and shifts and changes unpredictably and delightfully.

DonnyV 3y ago

That article was interesting but holy crap what was the deal with the constant shifting of the layout. Ads kept popping in and out shifting the content up and down. Its unreadable.

23B1 3y ago

I'm not worried in the medium-to-long term. Tech (which is nothing more than a lackey of Big Business) will never understand that real demand for cultural artifacts – art, books, film – comes from the consumers' need for real human connection. (The rare exception might be a guy like Steve Jobs). This is a huge advantage to artists and creative people because they can and will bypass that mindset entirely, and can do so with the confidence that they're connected to what human beings – consumers – actually want. It's why 'tech forward' orgs like NFLX can't really do much except churn out derivative works, meanwhile you've got the Davids, the Wachowskis, the Cohens, Tarantino, Chris Nolan, etc. constantly pushing the limits of creativity and writing their own ticket, without the permission of big business and its lackey tech bros. When these guys 'break out' of course everyone wants to be their best friend, but in no way was Warner Brothers or Paramount able to mechanize or catalyze that creative genius. What chaps my ass the most about the current crop of GAI is that its gone after writers and illustrators and artists because, in the process of sharing their works with the world, had that work hoovered up and used for training data without their permission or recompense. Luckily, all of the mediocrity these systems shit out is obviously superficial and soulless. Yeah you might some day be able to do a fully reactive murder mystery in Zuck's metaverse, but eventually you're going to want to be intrigued and delighted by something truly fresh and human – and that can only be created by other humans.

mjr00 3y ago

> A few days later I found myself in a grocery checkout line, skimming through yet another article in which the writer touted the wonders of “artificial intelligence” and fretted hazily over whether we are nearing the point when AI will be able to produce novels, films, and other creative work, effectively replacing us. When I looked up and over to other people in the line, half of them wore the same shoe brand as me. This is an interesting phenomenon with AI. People immediately assume that because AI has the ability to flood the market with mediocre art/writing/whatever, that artists and writers are screwed. In my experience it's quite the opposite; when the market gets flooded with repetitive mediocre garbage, the relative scarcity of something unique increases and actually increases the price, accordingly. To the author's point, you can get the same Adidas shoe as everyone else for $50, or you can get a largely similar-looking pair of Guccis for ~$800 US[0]. Those Guccis would not exist if not for the mass-market cheap option; there would be no reason to splurge on an expensive unique purchase if everyone else was already wearing something unique. Or not wearing shoes at all. This has been my experience since the introduction of stable diffusion and dall-e as well. Just anecdotal, as I don't have any access to hard data, but the artists I've spoken to say that their commissions and fanbox subscriptions have gone up since SD was released. Much like the trope of friend groups keeping an ugly friend around to make themselves look better, the mediocre and soulless AI art that's flooding the internet has, apparently, made people appreciate human output as a premium brand. [0] https://www.nordstrom.com/s/gucci-new-ace-sneaker-women/4701092 https://www.nordstrom.com/s/gucci-new-ace-sneaker-women/4701...

Workaccount2 3y ago
AI art isn't as good as human art. But in a few years it will probably be better. It's a foolish mistake to look at the state of AI today and say "Actually this is gonna be great for my creative business!" What is scary about AI is not ChatGPT or Midjourney, Dalle, whatever. What is scary is that they showed up 10 years sooner than the earliest predictions, and are still rapidly improving. That is what should give people pause and it's amazing how many people miss this.
- mjr00 3y ago
  > AI art isn't as good as human art. But in a few years it will probably be better. This is a very big "probably". About as big as the "in a few years, AI drivers will probably be better than human drivers" line from 2015. Regardless: an AI can definitively be better than a human at something like Go (a closed game with a clear win condition) or driving (a logistical task with no creativity). But how do you define "better" when it comes to art? From a technical perspective, AI is already better than the vast majority of human artists! Very few artists would be able to, say, draw images of nature to create something that has the same contours as a popular meme[0]. So, arguably, we're already at the point where AI-generated art is "better" than human art. The question is, does that matter? Or does it matter in the same way that it matters that a motorcycle is faster than Usain Bolt? [0] https://i.imgur.com/jsI0HZH.png https://i.imgur.com/jsI0HZH.png

Fripplebubby 3y ago

LLMs will predict the most likely output to the prompt input - does this mean that a hypothetical perfect LLM would produce the most generic and widely-acceptable response to any input, the "lowest common denominator" output? The author seems to think so, and indeed, with today's GPTs, that is often what you might get back. But I think there's a lot more going on here - with natural language, a lot of the training effort goes into producing natural language output which conforms to the formal rules of language (it is intelligible, syntactically correct or roughly so, follows grammar, and semantically "about" what it is supposed to be about). But the interesting thing about natural language is that there are infinitely many correct responses which are valid formal responses to the prompt, both ones that introduce new or rare creative possibilities and ones that tread the same ground as a wikipedia article, for example (which is usually intended to be very neutral). So an LLM can produce both types of responses - formally correct and boring, formally correct and interesting - and you can see this today by offering "mix-ins" to the prompt like, "Tell me about Saudi Arabia's ruling family, in the style of Peter Griffin". Of course, it won't select to use the style of Peter Griffin without that prompt, but I think you can imagine a "weirdo" LLM that is trained with some positive response toward creative or interesting output too.

Fripplebubby 3y ago
I don't mean to imply that creativity is something that can be quantified objectively (used as the objective function, even), but it is widely understood that LLMs can produce varied output and be given particular biases or personalities
hansvm 3y ago
LLMs don't generally predict the most likely output, except when people go in to tweak temperature and other settings to make them more predictable for certain more mechanical tasks. The whole point of an LLM is that its output is calibrated; it produces a text with the same conditional probability you would expect seeing that text in the wild (give or take that we still don't understand why neural networks generalize).
- Fripplebubby 3y ago

Workaccount2 3y ago

Are there gonna be enough hats for all these people to eat in a few years? I think the perpetual image in media that humans are a unique entity with a magical inner force has gotten to everyone's head. It's just a fiction guys. It strokes egos to sell copies. We are meat computers outputting computation. There is nothing special inside. On our current trajectory in a few years AI will be far more creative than any human. Then we'll have to trudge through grumblings about how it feels so "cold and inhuman" despite blind studies showing people greatly prefer AI works.

markwkw 3y ago
I think there is something special. Some special coupling of emotional states between us (manifests as empathy, feeling of community, feeling of connection). Not saying that this connection cannot be replicated or faked, this remains an open question. Perhaps building an AI vastly smarter than us will conflict with replicating this coupling. Perhaps we will prefer artwork from an AI with which we do feel this type of connection. Perhaps this will make us build an AI that is really close to us in terms of experiencing feelings an emotional states, even if it means it's less "smart". This wouldn't be a dull outcome...

lhnz 3y ago

Hm, my own intuition is that it'll raise the baseline experience but then cause hyper-competition by human creators to exploit its powers to the limits while infusing further human creativity on top of what is mechanistically produced. I do not think this will result in overall mediocrity. It'll lift those with no access to art or way to pay for human creativity away from absolute poverty, while allowing those that already had access to these things a more competitive market.

Raidion 3y ago

I've posted about this before, but this is just the hollowing out of the "middle" of the value curve. Most people want something that's good enough, and they're not willing to pay for something that's N+1 good at 2N price. Other people want something that's N+5 good and will pay 3N price. As AI pushes down the lower bound of cost for creativity, the "mediocre" content creators/programmers/writers are going to get squeezed. At the same time, this means there is an untapped market for the people who can truly be creative and do great work. Thinking the difference between the Sports Illustrated AI articles and premium sports journalism (The Athletic), blog spam recipes vs America's Test Kitchen, or CNN/Fox front page vs NYT/WaPo/Politico, etc. You're going to have to vote with your dollar to support the things you care about. If you're not willing to pay for the higher quality, you're going to get AI (or other "budget") content that's making it's money by just being good enough to get you to see the advertisements. It's a free market out there and you need to be aware that just a few percent of people (probably roughly similar to the lurker/poster ratios) are the ones that enable high quality content to be successful. The HackerNews crowd are significantly more likely to be able to actively support creative pursuits that they find valuable, so I'd suggest you do so! Otherwise, when we don't reward quality, why wouldn't people just be mediocre?

bbotond 3y ago

It's ironic how the blandest, most predictable opinion piece is about how AI produces medicore content.

lcuff 3y ago
Totally agree. Insipid. No real insights, nuances or ideas.

thefz 3y ago

Rise? With the advent of social media, we have already witnessed a sea of mediocrity flooding the internet.

garethsprice 3y ago

Mediocrity is a good thing in the context of commodity goods. To use the author's example, you CAN still get unique custom made shoes and still buy hand-grown heirloom tomatoes at the farmer's market. The shoes will cost you $1,000+ and you'll pay a couple of bucks a tomato - prices which adjusted for inflation haven't changed since such practice was the only way people got shoes and tomatoes, and they got a lot less of them. Mediocrity and mass-production have made shoes and tomatoes accessible to everyone at far lower cost, and are not inherently bad.

tim333 3y ago

AI may have the opposite effect to the "five cars identical in make, model, year, and color", "same shoe brand" stuff in that you'll be able to have it design something unique for you. Already you can do that with digital artwork and it may extend to customising physical objects soon.