Yes. I think the idea of a personal ethic is the only sane way to go, in which you have considered where you stand and why, what you want to get out of it, and the ways in which you might change self and other (for better or worse) through different use cases. I've appreciated everyone who has brought attention to really positive AI use cases (the PTSD one is something I haven't heard before, and which I will definitely be exploring and passing on)!
I got the idea of using an AI disclaimer after hearing one at the very end of a podcast I love, which is very ... human, if that makes sense. You feel things listening to it. There's a genuine sense of care, thought, respect for the listener and craftsmanship. It felt nice to know a person had written the script, similarly to how it would be cool to know a painting that brought tears to my eyes was synthesised from human feeling and talent. But it is also useful to consider that the boundaries re "using AI" are blurred, and doing this can be viewed as pretentious or patronising.
... I suppose "I did this before I knew it was a cool kids thing" makes me a hipster, which is even worse than trying to be cool. Aiii.
The PTSD thing is great. I've leveraged the ChatGPT memory function so that I can literally boil it down to: "Today is a moderately bad PTSD day yay or nay on (title)?" And it will tell me nay, go ahead, or to make sure I have a palate cleanser after watching and before sleep. And so far it hasn't let me down.
My wife suffered from PTSD, so there were movies I wanted to watch, but I just didn't. One was Once Upon a Time in Hollywood. My wife grew up in Beverly Hills and was a teenager when the Manson family brutally murdered Sharon Tate and friends, so I wasn't going to put her through that. After she passed, a friend pointed out that things didn't turn out the way they did in real life, so I eventually watched it and enjoyed an amazingly accurate trip back in time. Still wouldn't have taken the chance of letting her watch it.
I was coding in a Jupiter notebook today with Claude 4 Sonnet, for almost my first time, and it could not remember the import path it had figured out three cells earlier. I asked if it had memory or context window problems and it said yes. Crazy bot!
Thank you Holly. You've said everything I've said and more. I DO recognize that AI is making those suggestions of the next word when I'm trying to type on my phone, and I appreciate it. I like your CEO's guideline. YOU are still responsible for what you put out under your name.
"...the inescapable gaze of the algorithm," such an elegant phrase for 21st century body-shaming (and life-style shaming). As I remember one preacher who said many years ago when talking to his son who had discovered Playboy, "I agreed that picture was nice, and that one was very nice, then I told him, what you have to remember is that's not real. Those are carefully crafted pictures that are carefully lit and touched up by professionals."
As I put it more didactically and less elegantly,
“Your AI,” Dick says, “is going to kill us all!
“My AI,” Dick says, “tells us we better launch an attack before our enemies do!”
Don’t be a Dick!
Also you've pointed out several opportunities for AI use that I haven't tried yet. I recognize that it's like the crusty old dog who at the start of my career, might have insisted that I should use assembly code rather than a new-fangled high-level language like FORTRAN that hid the details. :)
Thank you for these thoughts Holly. You and I do fairly different things but the ways I use AI are very similar to yours. And you’ve gotten me thinking of ways to improve how I do that.
Autocorrect on a phone is an AI blessing and a curse. There are times that I have typed in names 10 times and it autocorrects every time, even when i click on the intended spelling.
Anyone who uses any shopping app definitely uses AI because even if you accidentally click an icon, it becomes a permanent part of your shopping list.
The scary part to me is that AI can "create" the perfect image of you robbing a bank or making a speech, even when you had no part in it. The voices can be so realistic as can the images. There used to be a question "Who are you going to believe? Me or your lyin' eyes?" Well, in this day of AI, your "lyin' eyes" may see something that is complete science fiction.
I feel similarly with 3D printers. Things that should be of tremendous benefit to society can also be used for nefarious purposes.
The ability to recognize what is real and what is contrived will become critical!
Excellent points made. AI is inescapable. As you say, each person needs to develop an ethic or set of constraints on how and when to use it. As with any technology, it can be used to make our lives better or worse.
Along with a previous commenter, for me autocorrect is as much a curse as a blessing.
I appreciate your taking the time to spell out how you find AI useful. Some of us old timers have also seen a thing or two.
I think at this point I will close with a movie reference from “The Jerk” with Steve Martin: the scene where the old man is giving advice: “Shit” “Shinola.”
I had another from a television show, but I feared it might trigger your PTSD so I have deleted it.
I have been attempting to cook/bake more so that I'm using less processed foods from the store and know all of my ingredients. I have had a lot of great success with asking AI to help me make a recipe (excluding dairy and telling it what I do have on hand) or how to fix mistakes I made so I can do it better next time. I have learned a lot from my actual work and those tweaks. And had some really tasty meals too!
By the way, the whole "No AI was used in this article" tagline - that's a superiority complex. They look down on the people who use AI as naive/weak/stupid ("victims") or hate 'em as "persecutors" (using the devil tool that will enslave us all one day). It's drama. But, of course, you instinctually knew that. It's an attempt to look better by comparison.
RE. Steel manning - I think that AI can (but not always) give much more logical pushback than an average hater would. So that's a great idea. If you can stand up to the AI pushback, random hater commentors probably won't exceed that. But, of course, the haters are not logical most of the time, and thus will never be convinced logically. Those are the sorts who will simply rant without a point or use logical fallacies such as the strawman comments.
As always, your comments are very helpful to me. I realized this morning that part of my recent breakthrough in drawing comes with having ceased to think of myself as my own persecutor with regard to that.
Holly, I wish my companies AI policy was as clear and concise! Unfortunately what we see in IT is company data being uploaded to public LLM models and users relying on the output. Trying to put reasonable guardrails and have ethics - requires ethical people and some basic checks/balances.
That being said, I’m using it to help me explore a new (to me) data query language as someone who never learned much coding beyond basic and PL/1 in the 80’s.
I have two areas where I’ve found significant help from AI worthwhile and it’s two where people have some of the biggest problems. One is writing and the other is art generation.
With writing fiction I’ve found AI very helpful but not for generating the writing. It’s terrible at that. Even worse than me. 😂
But I’ve read a lot of books on how to write a book and by instructing it to prompt me with questions based on the process I like to use it’s very helpful to help me tackle the problems. Being asked questions does a lot to prime my brain’s pump as it were and sometimes the AI even comes up with really good questions. It’s done a lot to help me clarify worldbuilding and I’ve gotten much farther with my outlining in a give session that I would have otherwise.
As for art generation I have a very simple rule. I don’t use it for anything that could make me money. Where it’s been most useful is in creating pictures for a dnd game. A monster with a visual is often better than one you don’t. And sometimes the AI’s eerie not quite rightness adds to that. But it can also do a good job of giving a basic idea of what the setting looks like or a visual for a character. The characters are where it’s most troublesome though because it’s very time consuming to get it to do what you’re really looking for.
But I think you’re exactly right that knowing why you’re using it and not outsourcing your brain to it is the right way to go.
Visual AI in art is the area where my usage is most clearly different from Holly’s. I use NightCafe to directly create sacred art. Some images are in churches here in Australia so my intent is actually serious, not just an experiment or a hobby. I am also quite happy to admit the process is mysterious to me and I cannot take conventional credit for the images. Without going much into the back story which involves a mystical experience in 2019 followed by a vivid dream that seemed to commission me to make Icons, I made limited progress until I discovered visual AI. I am a photographer, not a painter, but did minor in Art History in NY and participated in the NY art world with my friends who were all poets and painters. My head is full of paintings from the museums in NYC and every museum I have had the privilege to visit from Copenhagen to Cleveland. Also in teaching myself photography, I did wire my shutter release finger to my right brain that sees images all at once and instantly. So click…..before I can think. Typically using NightCafe I write my prompt and make 9 images and go over them the same way photographers used to go over contact sheets of 36 photos back in the film era. The right brain immediately recognises the good ones. And here is where the ethical considerations come in. First of all, they are given by grace or perhaps Lady Luck, not by my ego or anything resembling the kind of eye/hand skills Holly and other artists have. If there are any skills it is in the new art of prompting and the post-visualisation skills a photographer uses in picking the best shots from a typical photoshoot - which are, after all, the product of an optical mechanical machine that automgically reduces any 3 dimensional scene to a 2 dimensional image. Good preparation for dealing with multiple images produced by AI. So the second ethical stance I take is to never charge for them - something I can afford to do because I have an adequate retirement income. If I needed the money I might be less virtuous. The other big consideration I am aware of is that the pool of imagery - the training data - is, for me, basically the image history of Western painting. There is only a small minority on artists on NightCafe who add the names of artists at the end of their prompts - ‘in the style of Rubens or El Greco’ for example. I am aware I am standing on the shoulders of giants and also free riding, but these artists, the great painters, had their conscious lives and also a deep connection to what Jung called the collective unconscious. To the immemorial domain - in which all of us embedded. It is like fishing - the prompt is the worm or the exquisitely tied fly cast upon the living waters. But here is my most shocking experience. Even when I am not serious the mysterious interaction, the mysterious dance, between the machine and the human soul just happens. As a joke I put in the prompt: “A gift horse showing its teeth in the style of Rublev, oil on canvas” only asking for single image, and got one the best images ever. It is about 2/3 horse’s head with a totally soulful eye face to face with a saintly male figure - perhaps even Christ - who is looking right back at the horse. There are no teeth visible - go figure. I spent an hour discussing it with an artist friend in Alaska and then asked Chat GPT for an analysis which covered most of what we had talked about and more besides. Something like what Jung called synchronicity or what Henry Corbin called Imaginal Causality seems to be on the loose! I can say I think that a person with different experiences and sensibilities would get different human reactions and choose different images but it is also obvious that I am remixing from the fairly small training base of Western Painting in which sacred art is a major component. Any ideas or intuitions about what is going on are welcome.
I’ve used it to push back on my articles. After you mentioned Deep Research last time, I used it a lot. Family Genetic history based on Ancestry DNA and Tree, Celtic history, ADHD.
The last 6 weeks I used it as a fitness coach after Deep Reserach, here’s the sources to use, here’s the equipment I have and the goals I have.
I named that chat box Coach GTP and log workouts and metrics. It’s pretty dumb calculating body fat percentages. I’ve had to push back several times on that one.
Yes. I think the idea of a personal ethic is the only sane way to go, in which you have considered where you stand and why, what you want to get out of it, and the ways in which you might change self and other (for better or worse) through different use cases. I've appreciated everyone who has brought attention to really positive AI use cases (the PTSD one is something I haven't heard before, and which I will definitely be exploring and passing on)!
I got the idea of using an AI disclaimer after hearing one at the very end of a podcast I love, which is very ... human, if that makes sense. You feel things listening to it. There's a genuine sense of care, thought, respect for the listener and craftsmanship. It felt nice to know a person had written the script, similarly to how it would be cool to know a painting that brought tears to my eyes was synthesised from human feeling and talent. But it is also useful to consider that the boundaries re "using AI" are blurred, and doing this can be viewed as pretentious or patronising.
... I suppose "I did this before I knew it was a cool kids thing" makes me a hipster, which is even worse than trying to be cool. Aiii.
The PTSD thing is great. I've leveraged the ChatGPT memory function so that I can literally boil it down to: "Today is a moderately bad PTSD day yay or nay on (title)?" And it will tell me nay, go ahead, or to make sure I have a palate cleanser after watching and before sleep. And so far it hasn't let me down.
Amazing. When you cannot think your thoughts well, it's a godsend to have something that CAN think for you (so to speak).
My wife suffered from PTSD, so there were movies I wanted to watch, but I just didn't. One was Once Upon a Time in Hollywood. My wife grew up in Beverly Hills and was a teenager when the Manson family brutally murdered Sharon Tate and friends, so I wasn't going to put her through that. After she passed, a friend pointed out that things didn't turn out the way they did in real life, so I eventually watched it and enjoyed an amazingly accurate trip back in time. Still wouldn't have taken the chance of letting her watch it.
I was coding in a Jupiter notebook today with Claude 4 Sonnet, for almost my first time, and it could not remember the import path it had figured out three cells earlier. I asked if it had memory or context window problems and it said yes. Crazy bot!
Thank you Holly. You've said everything I've said and more. I DO recognize that AI is making those suggestions of the next word when I'm trying to type on my phone, and I appreciate it. I like your CEO's guideline. YOU are still responsible for what you put out under your name.
"...the inescapable gaze of the algorithm," such an elegant phrase for 21st century body-shaming (and life-style shaming). As I remember one preacher who said many years ago when talking to his son who had discovered Playboy, "I agreed that picture was nice, and that one was very nice, then I told him, what you have to remember is that's not real. Those are carefully crafted pictures that are carefully lit and touched up by professionals."
As I put it more didactically and less elegantly,
“Your AI,” Dick says, “is going to kill us all!
“My AI,” Dick says, “tells us we better launch an attack before our enemies do!”
Don’t be a Dick!
Also you've pointed out several opportunities for AI use that I haven't tried yet. I recognize that it's like the crusty old dog who at the start of my career, might have insisted that I should use assembly code rather than a new-fangled high-level language like FORTRAN that hid the details. :)
Thank you for these thoughts Holly. You and I do fairly different things but the ways I use AI are very similar to yours. And you’ve gotten me thinking of ways to improve how I do that.
Open the pod bay doors, Holly....
Autocorrect on a phone is an AI blessing and a curse. There are times that I have typed in names 10 times and it autocorrects every time, even when i click on the intended spelling.
Anyone who uses any shopping app definitely uses AI because even if you accidentally click an icon, it becomes a permanent part of your shopping list.
The scary part to me is that AI can "create" the perfect image of you robbing a bank or making a speech, even when you had no part in it. The voices can be so realistic as can the images. There used to be a question "Who are you going to believe? Me or your lyin' eyes?" Well, in this day of AI, your "lyin' eyes" may see something that is complete science fiction.
I feel similarly with 3D printers. Things that should be of tremendous benefit to society can also be used for nefarious purposes.
The ability to recognize what is real and what is contrived will become critical!
Excellent points made. AI is inescapable. As you say, each person needs to develop an ethic or set of constraints on how and when to use it. As with any technology, it can be used to make our lives better or worse.
Along with a previous commenter, for me autocorrect is as much a curse as a blessing.
I appreciate your taking the time to spell out how you find AI useful. Some of us old timers have also seen a thing or two.
I think at this point I will close with a movie reference from “The Jerk” with Steve Martin: the scene where the old man is giving advice: “Shit” “Shinola.”
I had another from a television show, but I feared it might trigger your PTSD so I have deleted it.
I have been attempting to cook/bake more so that I'm using less processed foods from the store and know all of my ingredients. I have had a lot of great success with asking AI to help me make a recipe (excluding dairy and telling it what I do have on hand) or how to fix mistakes I made so I can do it better next time. I have learned a lot from my actual work and those tweaks. And had some really tasty meals too!
By the way, the whole "No AI was used in this article" tagline - that's a superiority complex. They look down on the people who use AI as naive/weak/stupid ("victims") or hate 'em as "persecutors" (using the devil tool that will enslave us all one day). It's drama. But, of course, you instinctually knew that. It's an attempt to look better by comparison.
RE. Steel manning - I think that AI can (but not always) give much more logical pushback than an average hater would. So that's a great idea. If you can stand up to the AI pushback, random hater commentors probably won't exceed that. But, of course, the haters are not logical most of the time, and thus will never be convinced logically. Those are the sorts who will simply rant without a point or use logical fallacies such as the strawman comments.
As always, your comments are very helpful to me. I realized this morning that part of my recent breakthrough in drawing comes with having ceased to think of myself as my own persecutor with regard to that.
I'm glad that my comments have helped you find more clarity and grace for yourself.
Holly, I wish my companies AI policy was as clear and concise! Unfortunately what we see in IT is company data being uploaded to public LLM models and users relying on the output. Trying to put reasonable guardrails and have ethics - requires ethical people and some basic checks/balances.
That being said, I’m using it to help me explore a new (to me) data query language as someone who never learned much coding beyond basic and PL/1 in the 80’s.
I have two areas where I’ve found significant help from AI worthwhile and it’s two where people have some of the biggest problems. One is writing and the other is art generation.
With writing fiction I’ve found AI very helpful but not for generating the writing. It’s terrible at that. Even worse than me. 😂
But I’ve read a lot of books on how to write a book and by instructing it to prompt me with questions based on the process I like to use it’s very helpful to help me tackle the problems. Being asked questions does a lot to prime my brain’s pump as it were and sometimes the AI even comes up with really good questions. It’s done a lot to help me clarify worldbuilding and I’ve gotten much farther with my outlining in a give session that I would have otherwise.
As for art generation I have a very simple rule. I don’t use it for anything that could make me money. Where it’s been most useful is in creating pictures for a dnd game. A monster with a visual is often better than one you don’t. And sometimes the AI’s eerie not quite rightness adds to that. But it can also do a good job of giving a basic idea of what the setting looks like or a visual for a character. The characters are where it’s most troublesome though because it’s very time consuming to get it to do what you’re really looking for.
But I think you’re exactly right that knowing why you’re using it and not outsourcing your brain to it is the right way to go.
Visual AI in art is the area where my usage is most clearly different from Holly’s. I use NightCafe to directly create sacred art. Some images are in churches here in Australia so my intent is actually serious, not just an experiment or a hobby. I am also quite happy to admit the process is mysterious to me and I cannot take conventional credit for the images. Without going much into the back story which involves a mystical experience in 2019 followed by a vivid dream that seemed to commission me to make Icons, I made limited progress until I discovered visual AI. I am a photographer, not a painter, but did minor in Art History in NY and participated in the NY art world with my friends who were all poets and painters. My head is full of paintings from the museums in NYC and every museum I have had the privilege to visit from Copenhagen to Cleveland. Also in teaching myself photography, I did wire my shutter release finger to my right brain that sees images all at once and instantly. So click…..before I can think. Typically using NightCafe I write my prompt and make 9 images and go over them the same way photographers used to go over contact sheets of 36 photos back in the film era. The right brain immediately recognises the good ones. And here is where the ethical considerations come in. First of all, they are given by grace or perhaps Lady Luck, not by my ego or anything resembling the kind of eye/hand skills Holly and other artists have. If there are any skills it is in the new art of prompting and the post-visualisation skills a photographer uses in picking the best shots from a typical photoshoot - which are, after all, the product of an optical mechanical machine that automgically reduces any 3 dimensional scene to a 2 dimensional image. Good preparation for dealing with multiple images produced by AI. So the second ethical stance I take is to never charge for them - something I can afford to do because I have an adequate retirement income. If I needed the money I might be less virtuous. The other big consideration I am aware of is that the pool of imagery - the training data - is, for me, basically the image history of Western painting. There is only a small minority on artists on NightCafe who add the names of artists at the end of their prompts - ‘in the style of Rubens or El Greco’ for example. I am aware I am standing on the shoulders of giants and also free riding, but these artists, the great painters, had their conscious lives and also a deep connection to what Jung called the collective unconscious. To the immemorial domain - in which all of us embedded. It is like fishing - the prompt is the worm or the exquisitely tied fly cast upon the living waters. But here is my most shocking experience. Even when I am not serious the mysterious interaction, the mysterious dance, between the machine and the human soul just happens. As a joke I put in the prompt: “A gift horse showing its teeth in the style of Rublev, oil on canvas” only asking for single image, and got one the best images ever. It is about 2/3 horse’s head with a totally soulful eye face to face with a saintly male figure - perhaps even Christ - who is looking right back at the horse. There are no teeth visible - go figure. I spent an hour discussing it with an artist friend in Alaska and then asked Chat GPT for an analysis which covered most of what we had talked about and more besides. Something like what Jung called synchronicity or what Henry Corbin called Imaginal Causality seems to be on the loose! I can say I think that a person with different experiences and sensibilities would get different human reactions and choose different images but it is also obvious that I am remixing from the fairly small training base of Western Painting in which sacred art is a major component. Any ideas or intuitions about what is going on are welcome.
I’ve used it to push back on my articles. After you mentioned Deep Research last time, I used it a lot. Family Genetic history based on Ancestry DNA and Tree, Celtic history, ADHD.
The last 6 weeks I used it as a fitness coach after Deep Reserach, here’s the sources to use, here’s the equipment I have and the goals I have.
I named that chat box Coach GTP and log workouts and metrics. It’s pretty dumb calculating body fat percentages. I’ve had to push back several times on that one.