"But, like Guitar Hero, no outside creative skill has been acquired or developed."
Spoken like someone who never actually played the games. GH developed my interest in music as a kid, and Rock Band actually taught me how to drum in real life. It also garnered my interest in picking up and learning real guitar.
Those who forget that Rock Band not only a) existed and b) did everything GH did but a thousand times better + created by real musicians (Harmonix) do the genre, industry, and this article, injustice.
This article is written by someone stretching and reaching to relate two unrelated things. I'm offended by this.
The point being made isn't that it cannot inspire someone to then go on to learn but that mastery of the game alone does not convey creative skill. Of course someone can be inspired to go out and learn the skill being modeled but use of the model doesn't teach the skill even if it might make someone really excited to go learn.
Not the argument you're looking for, but I've been playing drums for over ten years now because of Rock Band. Even used the sticks that came with it on my first e-kit. The game didn't teach me how to drum, but it taught me that I had a much deeper interest in playing drums than I'd ever imagined. That led into years and years of practice and I've actually become pretty good.
And that scenario is covered in the post, in the seventh paragraph. Nowhere in it is Scalzi saying Guitar Hero can’t give you an interest to pick up the Guitar, quite the opposite. The point being made is that when you do pick up the Guitar, you’re starting to learn from scratch, your years of playing Guitar Hero don’t really give you a head-start.
As a person who plays several instruments (including drums) in professional situations, and as a person who enjoyed Guitar Hero, I always find it strange when folks are able to take the "experience" of Rock Band / Songstar /etc as a proxy for the experience of practicing music.
I actually do sight read for a couple of instruments- nothing in any of those games has literally anything to do with that.
Hell, my (very good) reading for clarinet and double bass has nothing to do with my (very poor) ability to sight read on piano.
Whatever gets folks interested in stuff is great, but to me it's like saying "I watched 10K hours of MTV in the 80s and that was very helpful in my career playing jazz"...
The same thing is true with writing and other skills replacements- much of my ability and desire to write comes from the experience of using it as an adjunct to thinking.
Having to figure out how I should connect ideas and state them in a way others will understand clarifies my own thoughts, so the idea that I am somehow gaining idunnowhat by adding a bunch of verbal noise to my writing misses a lot of the reasons why I like to write.
I'm happy to use LLMs to draft things like the contracts I need for the mechanical royalty payments I make when I record other folks material, but that's only because I am not trying to learn how to think legalistically.
And further, creating documents via might spark an interest in legal language and how it get used, but it's not going to help me understand the underlying reasons for what's going on any more than watching a whole lot of porn is going to help me be a better lover...
No true musician got their start in Guitar Hero, eh?
I got my start in drumming from playing on Rockband. When I mentioned it to the folk that I bought my first e-kit, they rolled their eyes. Fine, whatever, I expect musicians to be pretentious.
But yanno what? I still learned a lot. The controller itself was midi-capable so I was able to hook it up to my kit and play the game while practicing my form and timing. Was it perfect? Fuck no; it sucked. But it guided me into recognizing that I wasn't as good as I wanted to be, which led to lessons and many many hours of practicing.
I owe a lot to rock band because it opened a world of play beyond the game itself.
Sure, it’s all gamified and fun and even addictive, but at the end of the day no outside, transferable skill accrues to the player. To do that, you have to set the controller down, pick up an actual guitar, and start another, entirely different journey. And perhaps some Guitar Hero player did or will, which is awesome. It means at some point they were not satisfied with pretending to play the guitar, following someone else’s rules and set-up. They still started, effectively, at zero. There was a whole new skill tree to be learned that had very little to do with the game, outside of the shape of the controller.
I get the point, but I think it's not great. A lot of what's said here can also be said about cover bands or learning to play other peoples' music.
Scalzi has developed an almost religious sort of dogmatic hatred of AI. He seems to acknowledge none of the benefits, even those that are obvious. Likely because he feels it's harmful to the way he earns his living.
He is rapidly coming to define himself through opposition, rather than promotion. Which is a dangerous and fruitless game.
Throughout human history there has never been a still useful technology that we as a species chose to stop using, and AI very much is. Maybe we'll come to regulate it, but the genie will never go completely back into the bottle and folks are going to need to get used to the new world.
I use AI to code review my personal projects. There's no downside - I ignore what I don't like and take the suggestions that make sense. Most of the suggestions are good and it found a ton of little bugs over time.
I think that someone who's radically opposed to any use of AI is simply a poor developer.
You said it yourself - you have the knowledge and ability to know what to ignore and keep what makes sense, but that's because you spent years actually writing and understanding code.
The magic wand of AI is absolutely wreaking havoc with juniors engineers' ability to understand complex code bases and software engineering fundamentals. How will the engineers of tomorrow know what to keep and what to reject if all they've ever known is "you are a principal engineer. read this Jira ticket and build a solution. ultrathink".
What you said is different from being radically opposed to AI in all situations. You're saying that junior engineers should keep learning the fundamentals of software development instead of fully relying on AI. I agree with that. Inexperienced people shouldn't be left alone with dangerous tools, so to speak, but it doesn't mean that we should abandon these tools.
I think that junior engineers can learn from AI, but the corporate setting isn't the best place for self-learning from AI. There's a lot to be learned from the average advice from the better AIs, as long as they are trying to understand the feedback. Please note that for these engineers, AI is replacing the old copy-and-paste from Stack Overflow. At least with AI they can dig deeper and ask questions.
> Likely because he feels it's harmful to the way he earns his living.
Absurd. Scalzi fans are not going to stop reading his works because LLMs exist. If anything, established authors are the ones with the least to fear.
This talking point that when people dislike something it’s because it affects their personal finances is getting really tired. Not everyone is a greedy selfish prick who only thinks of themselves, people are allowed to dislike things for other reasons. That’s normal, it’s not an anomaly.
> but the genie will never go completely back into the bottle and folks are going to need to get used to the new world.
And apparently we’ll never share another original idea, either. “Genie is out of the bottle”, “you’re going to get left behind”. Apparently the people who called LLMs parrots got it wrong, they instead turn people into them, mindlessly repeating whatever the labs want them to.
Nobody even started to seriously write books using AI. But at one point there will be AI books on "most read" lists. And a more than one of those will have readership higher than anything by John Scalzi. And he's not bad, I liked what I read of his.
Plenty of people have been using AI to write books for a couple of years now.
The successful ones already knew how to write books - which directly contradicts the standard line about "It's just people who couldn't do it without AI."
That's just objectively wrong.
There have been some successful spammers who flood Amazon with tens/hundreds/thousands of titles, but that's more of an exercise in technology and marketing than writing.
Climate Change has accelerated rapidly. We're expecting 3 degrees of warming by 2050 now, 5 degrees by 2100. Betting on the singularity to save us is culty wankery.
DJing is an interesting art. It’s not creative per se, but can be done creatively. Modern DJ software+hardware in particular allows for a ton of on-the-fly remixing, clever use of fx and stems, and even some custom on-the-fly production setups. But the average DJ (and even electronic music producer) is much closer to Guitar Hero than they are to writing their own songs for guitar.
This is a bad analogy by somebody who never played the game. Guitar hero is not harder or easier than guitar, it's just different.
I don't know how to play guitar, yet I can play guitar hero. My friend who plays guitar cannot play guitar hero. I'm not interested in playing guitar and I don't find GH to be a "shortcut" to learn it. In fact, I play the piano and I'm okay like that.
Guitar Hero is a "rhythm game", perhaps calling it a "guitar simulator" is a bit incorrect. A rhythm game is one where you, for example, push buttons in tune with a song, exactly what Guitar hero is.
So you are good at different things. Maybe not at a real guitar but perhaps at reaction times.
I say this as someone who had one of those plug-in guitars but never knew how to use it, hah.
Bad analogy. In GH you have to press the buttons at the exact time that is prescribed and you're scored on that. With AI there is no such prescribed goal. The world is your oyster. You can use it in boring ways, but you can use it in immensely creative ways too. You can chain together AI agents, image generators, video generators, 3D asset generators, computer vision models, text embeddings etc. etc. and prompt all the orchestration into existence with your ideas, taste and direction. It's the opposite of a constrained game.
Boring people use it in boring ways, interesting people use it in interesting ways.
The rest is old man yells at cloud, like how people criticized "laptop music" or photography.
Except Guitar Hero is a game, and is marketed as such. Guitar Hero is like Bitcoin mining, you are doing useless computations to make some number go up, in the case of Bitcoin, your money, and Guitar Hero, your score.
But AI actually does something unlike Guitar Hero. There is a real output that can be used tomorrow even if AI "goes away".
You say AI helps you learn no skills, essentially. I don't know about you, but AI helped me understand Linux better in a way I didn't feel confident I would be able to before AI. I have created code with AI that I can maintain, and understand without it.
AI can go in conjunction with learning manually and can be helpful. Just because many examples exist of AI being unhelpful does not mean the whole technology is unhelpful.
Maybe not, but folks keep telling me that I'm ignorant for not agreeing that being able to "engineer" a prompt is the same exact thing as being a poet or painter.
All of my childhood friends who bragged about being good at GH/RB only ever bragged about being good at the _games_. Never ran in to someone trying to pass it off as being able to play the guitar.
Anecdata, friend. Two old friends became extraordinarily good at guitar hero and even won competitions. They're also some of the best guitar players I know. One of them is an extremely good drummer and can do improv piano like a beast. They'd both agree that guitar hero helped them play better.
What are your thoughts on games like Keyboard Hero?
I'm making a LED Camel Totem for Burning Man. I had the idea to make a Koons-style balloon animal.
I explored this idea with ChatGPT, asking it to make some reference images. I also asked for some prompt generation. I have weak sketching abilities (it's on my list to improve), so this is how I "storyboard".
To be cheeky, I asked Claude to make balloon simulator from which I could eventually drop into Fusion360 to muck with and 3D print. CamelKoons was born! [1]
I wanted a Golang backend and web frontend. Fable pretty much one-shotted the step-one scaffolding: Golang physical simulation of a single balloon, streaming geometry to a Web application. Vibed through various issues and then step two was multiple balloons with twists (which were just choke points).
In the world of disposable, hyper-focused software, this tool remained static to Dromedary Camel topologies. A generalized topology feature is left as an exercise for the reader's LLM.
I like empowering my agents to loop [2]. So, I used Claude-in-Chrome to do some loops where Claude positioned the Camel and could screenshot it. I also gave it an HTTP endpoint to position the camera for itself and me.
Eventually, I wanted to push views and comments from the Web page, so I added a comment box. Claude Code used its new-to-me "monitoring" feature to watch for that action and react to it in the session. Claude made itself a "look-at-camel" skill, so cute! [3]
While the balloons were looking interesting, I noted it wasn't very Koons-like and Claude reminded me that Koons was art and we were doing physical simulation. So then we started distorting the physical model. Of course we also added model export, etc.
After tokenmaxxing, I ended up with something that was a cross between a poop emoji and an alien dildo.
I'm now learning the Forms feature in Fusion360 and following a great balloon-dog tutorial on YouTube. I think this guitar approach will win out over my Guitar Hero approach.
That implies that you did more than prompt the AI and actually learned whatever activity it is you are asking the AI about, because otherwise you wouldn't know to ask those questions. A person making music solely with AI won't ask about tuning their guitar because they don't have one and don't intend to use one.
One could also make hotdogs and eat them without developing any skill as a chef, but it's unlikely there would be thousands of blog posts complaining about the downsides of microwaved hotdogs every week.
If they came out with nutri-paste that gives you all of your nutrition in a grey slurry drink, and replaced everything in grocery stores with it, people would complain
And presumably there would be similarly thousands of posts saying things like "I love my nutri-paste, it lets me skip the boring part of cooking my meals straight to the eating and getting calories and vitamins which is the real purpose of eating anyways"
I was trying to figure out what time of day, by UTC, would most people be asleep. Searching I could only pre-slopped answers in hundreds of random blogs.
So I decided to get my own custom slop. Opened Gemini and it said "video generation is free for a few days, try it out!". I uploaded a copy of Neuromancer and asked it to generate a scene.
MF had Cage inserting a power plug in his ear.
Ah and the time zone curiosity question? It went about it for a a minute or so and said "can't help you with that one, anything else?"
I think we've seen the best these LLMs can do and the race to the bottom has started. It's all cost cutting now, and I am very curious to see what will be of OpenAI and Anthropic. Probably will end up absorbed by MS, Amazon and Nvidia.
These folks are completely unaware that they have witnessed creative and high quality generative AI content without realizing what it was. Because that's the thing. Good generative AI is invisible per se. And it is a lot more ubiquitous than people think by now.
So high quality I can certainly acknowledge because indeed the output from these models can be indistinguishable from human generated content. However, I am curious to know what you mean by creative here, in two senses:
1) Do you have an example of AI generated content being particularly novel or different, and not just a (admittedly highly consistent) rehash of what it was trained on?
2) What do you think it means to describe the output of a model that doesn't possess any agency as "creative"?
I find the test of “show me AI art that isn’t a rehash” to be a bit pernicious because I have a hard time coming up with any recent examples of non-AI art that I could lie about and claim was AI generated and you’d respond “OK, that AI art is not a rehash.” If everything is a rehash, what are we arguing about?
“model that doesn't possess any agency” implies a model acting on its own. Like a paint brush flying around the room painting the walls with no people around. Or a computer spontaneously turning on and publishing images to an online gallery with no on operating it. The creative process of using AI tools is one of exploration, pipelining and editing.
Commentators strawman image generators as magic paintbrush machine guns. But, the process of using them intentionally is more like nature/street photography. You can’t walk into the forest and declare “I will photograph an owl fighting a fox over a mouse today.” But, you can develop a sense for where to go to find good shots. Now imagine instead of walking in a forest, you can pop anywhere in the multiverse. You don’t know exactly what to expect each time. And, the space is infinite in many dimensions. But, you can still learn to navigate.
From there you get into integrations with 2D painting/3D modeling tools and video editors. Generating references which are used to generate style sheets, then characters, sets, blockings, dialog, motion, base renders, upscales, edits.
“Make me a pretty picture. Masterpiece, best quality” is easy to make fun of. And, it’s common because so easy many people are playing around with it. But, it’s tiresome to find all of the discussion assumes that’s all there can ever be.
"But, like Guitar Hero, no outside creative skill has been acquired or developed."
Spoken like someone who never actually played the games. GH developed my interest in music as a kid, and Rock Band actually taught me how to drum in real life. It also garnered my interest in picking up and learning real guitar.
Those who forget that Rock Band not only a) existed and b) did everything GH did but a thousand times better + created by real musicians (Harmonix) do the genre, industry, and this article, injustice.
This article is written by someone stretching and reaching to relate two unrelated things. I'm offended by this.
The point being made isn't that it cannot inspire someone to then go on to learn but that mastery of the game alone does not convey creative skill. Of course someone can be inspired to go out and learn the skill being modeled but use of the model doesn't teach the skill even if it might make someone really excited to go learn.
Is anyone actually arguing that it conveys creative skill though?
Did you learn to play guitar playing guitar hero?
No, you didn't.
You're straw manning an argument that doesn't exist in the article
Not the argument you're looking for, but I've been playing drums for over ten years now because of Rock Band. Even used the sticks that came with it on my first e-kit. The game didn't teach me how to drum, but it taught me that I had a much deeper interest in playing drums than I'd ever imagined. That led into years and years of practice and I've actually become pretty good.
And that scenario is covered in the post, in the seventh paragraph. Nowhere in it is Scalzi saying Guitar Hero can’t give you an interest to pick up the Guitar, quite the opposite. The point being made is that when you do pick up the Guitar, you’re starting to learn from scratch, your years of playing Guitar Hero don’t really give you a head-start.
You've had years of listening to music and practicing a very rudimentary form of sight reading.
I'm not aware of any studies proving or disproving that this gives beginners a head start, but it would be an interesting thing to investigate.
A categorical "Nice try, but you've just been wasting your time" seems overconfident.
As a person who plays several instruments (including drums) in professional situations, and as a person who enjoyed Guitar Hero, I always find it strange when folks are able to take the "experience" of Rock Band / Songstar /etc as a proxy for the experience of practicing music.
I actually do sight read for a couple of instruments- nothing in any of those games has literally anything to do with that.
Hell, my (very good) reading for clarinet and double bass has nothing to do with my (very poor) ability to sight read on piano.
Whatever gets folks interested in stuff is great, but to me it's like saying "I watched 10K hours of MTV in the 80s and that was very helpful in my career playing jazz"...
The same thing is true with writing and other skills replacements- much of my ability and desire to write comes from the experience of using it as an adjunct to thinking.
Having to figure out how I should connect ideas and state them in a way others will understand clarifies my own thoughts, so the idea that I am somehow gaining idunnowhat by adding a bunch of verbal noise to my writing misses a lot of the reasons why I like to write.
I'm happy to use LLMs to draft things like the contracts I need for the mechanical royalty payments I make when I record other folks material, but that's only because I am not trying to learn how to think legalistically.
And further, creating documents via might spark an interest in legal language and how it get used, but it's not going to help me understand the underlying reasons for what's going on any more than watching a whole lot of porn is going to help me be a better lover...
No true musician got their start in Guitar Hero, eh?
I got my start in drumming from playing on Rockband. When I mentioned it to the folk that I bought my first e-kit, they rolled their eyes. Fine, whatever, I expect musicians to be pretentious.
But yanno what? I still learned a lot. The controller itself was midi-capable so I was able to hook it up to my kit and play the game while practicing my form and timing. Was it perfect? Fuck no; it sucked. But it guided me into recognizing that I wasn't as good as I wanted to be, which led to lessons and many many hours of practicing.
I owe a lot to rock band because it opened a world of play beyond the game itself.
There are people here who have been using LLMs for years.
What is that?
Arrested development?
The paragraph in-question:
I get the point, but I think it's not great. A lot of what's said here can also be said about cover bands or learning to play other peoples' music.Scalzi has developed an almost religious sort of dogmatic hatred of AI. He seems to acknowledge none of the benefits, even those that are obvious. Likely because he feels it's harmful to the way he earns his living.
He is rapidly coming to define himself through opposition, rather than promotion. Which is a dangerous and fruitless game.
Throughout human history there has never been a still useful technology that we as a species chose to stop using, and AI very much is. Maybe we'll come to regulate it, but the genie will never go completely back into the bottle and folks are going to need to get used to the new world.
I use AI to code review my personal projects. There's no downside - I ignore what I don't like and take the suggestions that make sense. Most of the suggestions are good and it found a ton of little bugs over time.
I think that someone who's radically opposed to any use of AI is simply a poor developer.
There's no downside for you.
You said it yourself - you have the knowledge and ability to know what to ignore and keep what makes sense, but that's because you spent years actually writing and understanding code.
The magic wand of AI is absolutely wreaking havoc with juniors engineers' ability to understand complex code bases and software engineering fundamentals. How will the engineers of tomorrow know what to keep and what to reject if all they've ever known is "you are a principal engineer. read this Jira ticket and build a solution. ultrathink".
What you said is different from being radically opposed to AI in all situations. You're saying that junior engineers should keep learning the fundamentals of software development instead of fully relying on AI. I agree with that. Inexperienced people shouldn't be left alone with dangerous tools, so to speak, but it doesn't mean that we should abandon these tools.
I think that junior engineers can learn from AI, but the corporate setting isn't the best place for self-learning from AI. There's a lot to be learned from the average advice from the better AIs, as long as they are trying to understand the feedback. Please note that for these engineers, AI is replacing the old copy-and-paste from Stack Overflow. At least with AI they can dig deeper and ask questions.
> Likely because he feels it's harmful to the way he earns his living.
Absurd. Scalzi fans are not going to stop reading his works because LLMs exist. If anything, established authors are the ones with the least to fear.
This talking point that when people dislike something it’s because it affects their personal finances is getting really tired. Not everyone is a greedy selfish prick who only thinks of themselves, people are allowed to dislike things for other reasons. That’s normal, it’s not an anomaly.
> but the genie will never go completely back into the bottle and folks are going to need to get used to the new world.
And apparently we’ll never share another original idea, either. “Genie is out of the bottle”, “you’re going to get left behind”. Apparently the people who called LLMs parrots got it wrong, they instead turn people into them, mindlessly repeating whatever the labs want them to.
> Apparently the people who called LLMs parrots got it wrong, they instead turn people into them, mindlessly repeating whatever the labs want them to.
Being rude about it doesn't make you right about it.
It sounds like people are repeating themselves because they have to keep repeating the obvious in every one of these threads:
The tools are useful, people are going to use them. Cope.
Nobody even started to seriously write books using AI. But at one point there will be AI books on "most read" lists. And a more than one of those will have readership higher than anything by John Scalzi. And he's not bad, I liked what I read of his.
Plenty of people have been using AI to write books for a couple of years now.
The successful ones already knew how to write books - which directly contradicts the standard line about "It's just people who couldn't do it without AI."
That's just objectively wrong.
There have been some successful spammers who flood Amazon with tens/hundreds/thousands of titles, but that's more of an exercise in technology and marketing than writing.
You're probably right, and it will be an incredibly sad day when we have completed replaced the human part of creation in our culture with AI creation
Might just be final nail in the coffin for our cultural stagnation
Climate Change has accelerated rapidly. We're expecting 3 degrees of warming by 2050 now, 5 degrees by 2100. Betting on the singularity to save us is culty wankery.
I doubt that any reasonable person is actually counting on that.
It's a new, and potentially very lucrative, technology. Given time the cost will go down, not up. The way that will happen is increased efficiency.
Do I know HOW that will happen? No, but I know that it will. It always has, and there are huge motivations for making it happen.
The most reasonable and sober-minded people are counting on that:
https://www.dpg-physik.de/veroeffentlichungen/publikationen/...
As a code monkey generative ai helps grow my skills daily, and all I have to do is read the code.
And to keep fluffing the control narrative, 0 mention of open weight models were made.
I don't know enough about DJing, but maybe that is also an analogy to GenAI usage that can work.
DJing is an interesting art. It’s not creative per se, but can be done creatively. Modern DJ software+hardware in particular allows for a ton of on-the-fly remixing, clever use of fx and stems, and even some custom on-the-fly production setups. But the average DJ (and even electronic music producer) is much closer to Guitar Hero than they are to writing their own songs for guitar.
This is a bad analogy by somebody who never played the game. Guitar hero is not harder or easier than guitar, it's just different.
I don't know how to play guitar, yet I can play guitar hero. My friend who plays guitar cannot play guitar hero. I'm not interested in playing guitar and I don't find GH to be a "shortcut" to learn it. In fact, I play the piano and I'm okay like that.
Guitar Hero is a "rhythm game", perhaps calling it a "guitar simulator" is a bit incorrect. A rhythm game is one where you, for example, push buttons in tune with a song, exactly what Guitar hero is.
So you are good at different things. Maybe not at a real guitar but perhaps at reaction times.
I say this as someone who had one of those plug-in guitars but never knew how to use it, hah.
He's not saying playing Guitar Hero is worthless. He's saying it won't teach you guitar, which is 100% true
So, given this is all obvious, the article is basically saying nothing.
I don't know how to tell you this, but the article isn't really about Guitar Hero
> I don’t know how to play guitar.
> Guitar hero is not harder or easier than guitar.
Ladies and gents, peak dunning krueger moment.
Bad analogy. In GH you have to press the buttons at the exact time that is prescribed and you're scored on that. With AI there is no such prescribed goal. The world is your oyster. You can use it in boring ways, but you can use it in immensely creative ways too. You can chain together AI agents, image generators, video generators, 3D asset generators, computer vision models, text embeddings etc. etc. and prompt all the orchestration into existence with your ideas, taste and direction. It's the opposite of a constrained game.
Boring people use it in boring ways, interesting people use it in interesting ways.
The rest is old man yells at cloud, like how people criticized "laptop music" or photography.
Except Guitar Hero is a game, and is marketed as such. Guitar Hero is like Bitcoin mining, you are doing useless computations to make some number go up, in the case of Bitcoin, your money, and Guitar Hero, your score.
But AI actually does something unlike Guitar Hero. There is a real output that can be used tomorrow even if AI "goes away".
You say AI helps you learn no skills, essentially. I don't know about you, but AI helped me understand Linux better in a way I didn't feel confident I would be able to before AI. I have created code with AI that I can maintain, and understand without it.
AI can go in conjunction with learning manually and can be helpful. Just because many examples exist of AI being unhelpful does not mean the whole technology is unhelpful.
“ Being a Guitar Hero player did not — and does not! — equal being a guitar player.”
Has anyone ever claimed otherwise?
Maybe not, but folks keep telling me that I'm ignorant for not agreeing that being able to "engineer" a prompt is the same exact thing as being a poet or painter.
All of my childhood friends who bragged about being good at GH/RB only ever bragged about being good at the _games_. Never ran in to someone trying to pass it off as being able to play the guitar.
Anecdata, friend. Two old friends became extraordinarily good at guitar hero and even won competitions. They're also some of the best guitar players I know. One of them is an extremely good drummer and can do improv piano like a beast. They'd both agree that guitar hero helped them play better.
What are your thoughts on games like Keyboard Hero?
Calculators have never improved my ability to do long division.
And yet, I am still quite good at mathematics.
I'm making a LED Camel Totem for Burning Man. I had the idea to make a Koons-style balloon animal.
I explored this idea with ChatGPT, asking it to make some reference images. I also asked for some prompt generation. I have weak sketching abilities (it's on my list to improve), so this is how I "storyboard".
To be cheeky, I asked Claude to make balloon simulator from which I could eventually drop into Fusion360 to muck with and 3D print. CamelKoons was born! [1]
I wanted a Golang backend and web frontend. Fable pretty much one-shotted the step-one scaffolding: Golang physical simulation of a single balloon, streaming geometry to a Web application. Vibed through various issues and then step two was multiple balloons with twists (which were just choke points).
In the world of disposable, hyper-focused software, this tool remained static to Dromedary Camel topologies. A generalized topology feature is left as an exercise for the reader's LLM.
I like empowering my agents to loop [2]. So, I used Claude-in-Chrome to do some loops where Claude positioned the Camel and could screenshot it. I also gave it an HTTP endpoint to position the camera for itself and me.
Eventually, I wanted to push views and comments from the Web page, so I added a comment box. Claude Code used its new-to-me "monitoring" feature to watch for that action and react to it in the session. Claude made itself a "look-at-camel" skill, so cute! [3]
While the balloons were looking interesting, I noted it wasn't very Koons-like and Claude reminded me that Koons was art and we were doing physical simulation. So then we started distorting the physical model. Of course we also added model export, etc.
After tokenmaxxing, I ended up with something that was a cross between a poop emoji and an alien dildo.
I'm now learning the Forms feature in Fusion360 and following a great balloon-dog tutorial on YouTube. I think this guitar approach will win out over my Guitar Hero approach.
[1] https://github.com/cameltopia/CamelKoons
[2] https://github.com/ConAcademy/WeaselToonCadova
[3] https://github.com/Cameltopia/CamelKoons/blob/main/.claude/s...
This seems disingenuous. I can ask a generative AI "explain this ternary operator to me" or "walk me through why this causes a race condition"
I cannot ask the video game why my real guitar is out of tune, or diagnose why I'm holding the fret wrong.
That implies that you did more than prompt the AI and actually learned whatever activity it is you are asking the AI about, because otherwise you wouldn't know to ask those questions. A person making music solely with AI won't ask about tuning their guitar because they don't have one and don't intend to use one.
One could also make hotdogs and eat them without developing any skill as a chef, but it's unlikely there would be thousands of blog posts complaining about the downsides of microwaved hotdogs every week.
If they came out with nutri-paste that gives you all of your nutrition in a grey slurry drink, and replaced everything in grocery stores with it, people would complain
And presumably there would be similarly thousands of posts saying things like "I love my nutri-paste, it lets me skip the boring part of cooking my meals straight to the eating and getting calories and vitamins which is the real purpose of eating anyways"
None of that has anything to do with creativity, which is what the post is about.
If you press the button on a camera, you are not a painter.
Don't worry, there's still room for paintings in human culture, even though photos outnumber them billion to one.
I was trying to figure out what time of day, by UTC, would most people be asleep. Searching I could only pre-slopped answers in hundreds of random blogs.
So I decided to get my own custom slop. Opened Gemini and it said "video generation is free for a few days, try it out!". I uploaded a copy of Neuromancer and asked it to generate a scene.
MF had Cage inserting a power plug in his ear.
Ah and the time zone curiosity question? It went about it for a a minute or so and said "can't help you with that one, anything else?"
I think we've seen the best these LLMs can do and the race to the bottom has started. It's all cost cutting now, and I am very curious to see what will be of OpenAI and Anthropic. Probably will end up absorbed by MS, Amazon and Nvidia.
You are bad at it. Don't get mad, get good.
edit: https://share.gemini.google/hvH2A3Yva5Uj
First try.
Wow, you typed out their question first try? What skill!
I doubt their attempt failed because they wrote the question poorly. It was probably just the service being flaky.
These folks are completely unaware that they have witnessed creative and high quality generative AI content without realizing what it was. Because that's the thing. Good generative AI is invisible per se. And it is a lot more ubiquitous than people think by now.
> creative and high quality generative AI content
So high quality I can certainly acknowledge because indeed the output from these models can be indistinguishable from human generated content. However, I am curious to know what you mean by creative here, in two senses:
1) Do you have an example of AI generated content being particularly novel or different, and not just a (admittedly highly consistent) rehash of what it was trained on? 2) What do you think it means to describe the output of a model that doesn't possess any agency as "creative"?
I find the test of “show me AI art that isn’t a rehash” to be a bit pernicious because I have a hard time coming up with any recent examples of non-AI art that I could lie about and claim was AI generated and you’d respond “OK, that AI art is not a rehash.” If everything is a rehash, what are we arguing about?
Instead, my goto examples of high quality creative works that are clearly grounded on AI tools is https://andreaciulu.com/projects/ and https://youtu.be/fyZhC2TXgcs
“model that doesn't possess any agency” implies a model acting on its own. Like a paint brush flying around the room painting the walls with no people around. Or a computer spontaneously turning on and publishing images to an online gallery with no on operating it. The creative process of using AI tools is one of exploration, pipelining and editing.
Commentators strawman image generators as magic paintbrush machine guns. But, the process of using them intentionally is more like nature/street photography. You can’t walk into the forest and declare “I will photograph an owl fighting a fox over a mouse today.” But, you can develop a sense for where to go to find good shots. Now imagine instead of walking in a forest, you can pop anywhere in the multiverse. You don’t know exactly what to expect each time. And, the space is infinite in many dimensions. But, you can still learn to navigate.
And, that’s just the process of direct prompting. One step past that gets you configuring pipelines of highly varied generative editing tools like https://www.runcomfy.com/comfyui-workflows/comfyui-product-r...
From there you get into integrations with 2D painting/3D modeling tools and video editors. Generating references which are used to generate style sheets, then characters, sets, blockings, dialog, motion, base renders, upscales, edits.
“Make me a pretty picture. Masterpiece, best quality” is easy to make fun of. And, it’s common because so easy many people are playing around with it. But, it’s tiresome to find all of the discussion assumes that’s all there can ever be.
Like what? Can you give me an example?