May 23, 2023·edited May 23, 2023Liked by Mark Palmer
(Going to try my best not to turn this comment into its own post lol)
Thank you Mark for opening this up for discussion, and doing so in such a welcoming way. Enjoyed reading this piece.
First and foremost, the images that generative AI is capable of can be truly stunning. No arguments there. The ease with which these images can now be generated is equally stunning. No argument there either.
Does this empower more people with more ways to produce artistic images? Absolutely. Does this bestow artistic genius on people who've never picked up a paintbrush? Absolutely not.
You ask "Was it 'art' to create these images?" The images themselves are art, yes. It's not human-made art, it's machine-generated art with input or creative direction from a human. If we must anthropomorphize, the AI is the artist and the human typing in the prompts is the client (not necessarily the creative director, you'd still need some professional cred for that)
I was intrigued by this paragraph: "If you discuss AI more than you use it, stop doing that. That’s what Kelly calls 'thinkism.' Only action leads to learning, understanding, and progress." So, for full transparency, I've used both generative image and text AI in my research. After all, I can't write about something I haven't actually used and worked with, to your point Mark. But because design, photography, and writing are in my professional wheelhouse (I'm a working author, amateur photographer, and I used to run a small design firm), I became pretty disenchanted pretty quickly—not so much bc of the quality (some of that yes) but more bc of the EXPERIENCE.
Here's the problem. Generative AI takes all the joy and challenge out of the process, the journey of artistic creation. No way in hell will I let ChatGPT write my posts or novels, because a) it's not me; b) I hate babysiting; and c) I enjoy, very much, the work of writing. So become the nanny of a robot writer I shall not :) Nor to feed the LLMs with my future writing do I wish.
This is not to say others should never use it. I can see plenty of use cases where gen AI is quite helpful and valuable.
Re: the painters revolt against the big bad camera and the new art of photography. The same can of course be said about digital tools like Photoshop or Painter. The difference with generative image AI systems is that they were trained on artists' work, without their knowledge, without their permission, and without compensation. The camera is a piece of mechanical equipment (yes now with software)—not trained on artists' work. Digital tools are software also not trained on artists' work. This is where many artists and writers and other creators feel the betrayal has taken place—it is not the algorithm itself but the way in which the developers chose, consciously, to train it. They could have contacted the artists and photographers beforehand. You know, worked with them, asked for permission. Would they have been able to assemble training data sets as large as the ones they've got? Probably not. This is where these thorny ethical questions lie. This many people feeling this passionately about a technology can't be wrong—and they're not wrong. This is their passion, their right, their livelihood in many ways.
One more thing and I'll hop off the pedestal. At the end of the day, we need to ask ourselves, how much of our creative agency and our own intellect do we want to farm out to LLMs, and at what price? It's an open-ended question.
Mark, I love this post. I've been guilty about talking AI more than working with AI. I'm using this as inspiration for our agency strategy...how do we embrace AI more fully?
In the comments on "Thinkism," Zan Tafakari referenced Paul Skallas' writing on this topic, which I had not read before. I found it; indeed, it's terrific and aligned with this article. Here it is, well worth a read:
For those not familiar with the term, anthropomorphizing is the act of "attributing human-like characteristics, emotions, and traits to non-human entities, such as animals, objects, or even abstract concepts. This can involve giving them human-like names, assuming they have intentions and motivations similar to humans, or interpreting their behaviors and reactions in a human-like way."
I tried to guard against anthropological thinking by choosing a camera as the closest cousin to generative AI, and thinking about it that way. I guess an intern would be in the danger zone, but that's simply to try to show the interaction model or relationship with AI -- that you have manage, edit, judge, ignore, and build upon its output.
But it's a fantastic point and a great 25-cent word to know!
Here's an article from Thabata Romanowski about a data + analog art project -- she started with DATA about her world and created ANALOG art from it.
I found it fascinating to read in the context of "what would be the role of AI in her process?" and thought I'd share it here. It's a great read:
https://nightingaledvs.com/data-art-self-portrait
(Going to try my best not to turn this comment into its own post lol)
Thank you Mark for opening this up for discussion, and doing so in such a welcoming way. Enjoyed reading this piece.
First and foremost, the images that generative AI is capable of can be truly stunning. No arguments there. The ease with which these images can now be generated is equally stunning. No argument there either.
Does this empower more people with more ways to produce artistic images? Absolutely. Does this bestow artistic genius on people who've never picked up a paintbrush? Absolutely not.
You ask "Was it 'art' to create these images?" The images themselves are art, yes. It's not human-made art, it's machine-generated art with input or creative direction from a human. If we must anthropomorphize, the AI is the artist and the human typing in the prompts is the client (not necessarily the creative director, you'd still need some professional cred for that)
I was intrigued by this paragraph: "If you discuss AI more than you use it, stop doing that. That’s what Kelly calls 'thinkism.' Only action leads to learning, understanding, and progress." So, for full transparency, I've used both generative image and text AI in my research. After all, I can't write about something I haven't actually used and worked with, to your point Mark. But because design, photography, and writing are in my professional wheelhouse (I'm a working author, amateur photographer, and I used to run a small design firm), I became pretty disenchanted pretty quickly—not so much bc of the quality (some of that yes) but more bc of the EXPERIENCE.
Here's the problem. Generative AI takes all the joy and challenge out of the process, the journey of artistic creation. No way in hell will I let ChatGPT write my posts or novels, because a) it's not me; b) I hate babysiting; and c) I enjoy, very much, the work of writing. So become the nanny of a robot writer I shall not :) Nor to feed the LLMs with my future writing do I wish.
This is not to say others should never use it. I can see plenty of use cases where gen AI is quite helpful and valuable.
Re: the painters revolt against the big bad camera and the new art of photography. The same can of course be said about digital tools like Photoshop or Painter. The difference with generative image AI systems is that they were trained on artists' work, without their knowledge, without their permission, and without compensation. The camera is a piece of mechanical equipment (yes now with software)—not trained on artists' work. Digital tools are software also not trained on artists' work. This is where many artists and writers and other creators feel the betrayal has taken place—it is not the algorithm itself but the way in which the developers chose, consciously, to train it. They could have contacted the artists and photographers beforehand. You know, worked with them, asked for permission. Would they have been able to assemble training data sets as large as the ones they've got? Probably not. This is where these thorny ethical questions lie. This many people feeling this passionately about a technology can't be wrong—and they're not wrong. This is their passion, their right, their livelihood in many ways.
One more thing and I'll hop off the pedestal. At the end of the day, we need to ask ourselves, how much of our creative agency and our own intellect do we want to farm out to LLMs, and at what price? It's an open-ended question.
(p.s. Ray you can stop with the pills now really)
Great essay. The only thing I'd gently caveat is to do this things but to be careful about anthropomorphizing.
Mark, I love this post. I've been guilty about talking AI more than working with AI. I'm using this as inspiration for our agency strategy...how do we embrace AI more fully?
In the comments on "Thinkism," Zan Tafakari referenced Paul Skallas' writing on this topic, which I had not read before. I found it; indeed, it's terrific and aligned with this article. Here it is, well worth a read:
https://paulbakaus.com/ais-biggest-bottleneck-in-2023-humans/
For more on this topic, read Thinkism, part one and two:
https://technosapien.substack.com/p/thinkism-part-1
https://technosapien.substack.com/p/how-to-beat-ai-thinkism-part-2
Thank you Michael!
For those not familiar with the term, anthropomorphizing is the act of "attributing human-like characteristics, emotions, and traits to non-human entities, such as animals, objects, or even abstract concepts. This can involve giving them human-like names, assuming they have intentions and motivations similar to humans, or interpreting their behaviors and reactions in a human-like way."
I tried to guard against anthropological thinking by choosing a camera as the closest cousin to generative AI, and thinking about it that way. I guess an intern would be in the danger zone, but that's simply to try to show the interaction model or relationship with AI -- that you have manage, edit, judge, ignore, and build upon its output.
But it's a fantastic point and a great 25-cent word to know!