No kidding — I continue to be gobsmacked by this rhetoric. Cheating is not only bad; it's a fakery of self, whether students (or any of us) do it for nefarious reasons or out of desperation. The last thing we need is more fakery in digital communications.
Indeed, it's already generating lawsuits, including with the New York Times. I expect that one will settle (with lots of posturing on both sides), but there's no way around the ethical issue of writers' work being scraped and dumped with no credit into a gigantic AI system — and, to the practical point, no recompense for the work done.
Have you seen the Poetry Camera? Point it at something and instead of a photo it spits out an AI-generated poem that captures (?) the image in verse. Chatbots are the still-distant but audible death knell for creative writing.
I’ve heard about the Poetry Camera - I’m not sure what I think of it yet (I’d need to test it), but my initial thought is that we (humans) get to decide what creative writing is, how to judge it, how to make meaning of our experiences. I think one thing many literary writers don’t get about AI is how good it its at divergent thinking - putting random stuff together - one of the hallmarks of creativity. But that creative randomness has to move us, too.
I don't feel that I am a writer and I have received no training. I just go by "what sounds good" to my ear. My interest is more on the development of ideas, new to me, and not on the flow of the English language. I do try to express those ideas without confusion about my intended meaning. So for me, the process of writing is where the learning happens, (just as you said).
We can consider what is the purpose of that writing, and can ChatGPT fulfill that purpose. ChatGPT can probably be entertaining, and maybe even humorous. Maybe you can sell it, in some market or other, and with or without acknowledging your sources or honesty about what you know and don’t know.
Whatever the purpose of writing (to sell something), the underlying purpose of all exploration is to get a grip on life. That means to increase your ability and confidence quotients and to decrease your victim quotient. ChatGPT will not fulfill that purpose for writers. And frankly, I don't think you can increase confidence by merely reading, (anything). Confidence is about experience, which is doing something, not just mulling ideas. If you are not prone to do something about, (experiment with) what you read, I say it is useless.
When reading about ChatGPT, Google Gemini, Meta Llama2, and the like, they are all Large Language Models. All LLM’s are based only on words. Their data is ALL the published digital written work. It seems to be a lot. Then something in the algorithm throws out inconsistencies and what they don’t like. It is said that there will never (never) be an unbiased LLM. A breakthrough beyond pre-published words is not on the horizon, this said just recently by Yann Lecun, the LLM developer for Meta.
I have long ago determined that the Big Tech Business Plan is not in advertising. Have you ever craved something or bought anything because of a computer pop-up screen? I honestly can't even remember one ad that was shown to me in 20 years. Who was the company, what was the product, what was the offer? Blank Slate.
Ads are just mini-computer programs inserted onto your page that monitor your clicks and searches. Anyway I always use one of the many privacy browsers that block ads. So the real business plan of Big Tech is to collect DATA and IDENTITY. We all know they do that, enough to channel our viewing and searching experience. I am not going to expand on that, since it is way more than any of us could believe possible.
So what is the next step in Data, and Identity collection? This is where the LLM comes into play. Now you will be taught to voluntarily "line-up" to submit your every thought, and every movement to Big Brother (Sam Altman). No need for spies, since you are always spying on yourself. Data collection will go up by orders of magnitude, and the existing Big Tech collections of Apple, Google, Facebook, Amazon, and Microsoft will wither and die. (Of course they will all develop LLM's too).
Why, you are preaching to the converted - I’m trying to give a tech CEO like Altman the benefit of the doubt in terms of having some sense of responsibility, but really, they want our data and will monetize it any way they can. That’s why their earnest statements about making the future a better place are so vague. These are smart people, and they could be less vague. They choose to keep what they’re really doing obscured, because that would not make them seem heroic.
I just finished reading a book by Debora Tannen (78), professor of linguistics at Georgetown University, do you know her? "Talking from 9 to 5" (2001) is about conversational styles in the work environment, and the conversational "rituals" that are used in these styles. Much of it exposes the tendencies of male and female coworkers, and what communication works for them, and what doesn't, (and why). She has written maybe a dozen books, but I think this one encapsulates much of her knowledge.
This book answered so many of my questions and brought new insight into people's understanding and misunderstanding, and into clear writing. It even allowed me to better understand you, Martha, and then to understand me, in how I feel to respond in our written exchanges.
In her research she was privileged to sit in on 100's of corporate settings and take notes on, or record conversations in business meetings. She then analyzed the communications styles. Some of her examples would be of interest in the university setting, from the trustee meetings, to departmental meetings and also to student groups as they worked through their communication assignments. Other settings were in the publishing industry, when they hire female authors to attract more female readers, but then the male editors squash their topic suggestions. In hospitals, with lawyers, in sales and marketing, if you haven't read it I'll give you my copy:
My proposal is to ask Chat GTP for a 2,000 word book review of "Talking from 9 to 5". Then you could paste it here, and we will see how well it captures the book. Or you can author a new post with it.
Why, I'm just catching up now with some comments (I'm traveling, and somehow I missed this one). I do know Tannen's work, and I'll say more about that on your site. The gender binary never rang true for me in her analysis, but I always liked the research she did on different linguistic styles and ethnic identity, such as the "engaged" style that I picked up from my Italian-American mother :-)
Just now I posted a second segment on Tannen. What do you see as "true" on these 8 rituals? I don't read a "hard gender binary" in her book. (I have only read one of them.)
I have to say; with these insights, I better understand you, and how you respond to me and your other commenters. And; I better understand me, and how a feeling to respond in a certain way "comes up". With this new understanding I am better at syncing between us. So Tannen has made a great difference for me.
First, Substack may be the least objective place on the Internet for discussions of AI content, due to the "writer as hero" culture the platform is built around.
Next, critics of the quality of AI content are typically not really that interested in the quality content issue, or they'd complaining about the extreme low quality of social media platforms, including Notes. If they were sincerely interested in quality content, they'd be looking for ways to improve the quality of content on social media, and would be embracing AI's ability to elevate the quality of that content in an extremely convenient, typically free, and highly scalable manner.
I've yet to see anyone on Substack be honest enough to simply admit that when it comes to social media, ChatGPT is a far better poster than most of us most of the time. ChatGPT isn't the problem, we are. The reason that ChatGPT can easily out perform we humans in social media most of the time is that we human WRITERS tend to engage on social media in a consistently lazy and self obsessed manner.
So many writers feel they are being demoted by AI, when the opposite is true. AI offers us the opportunity to take on a role that might be compared to the director of a movie. As a director, our job is to have the overall vision for the project, a higher order of thinking than just arranging words in a pleasing manner.
Or, if you prefer, think of an editor at the New York Times. The editor is a higher ranking higher paid member of the team, because they do the higher level big picture thinking. And the editors have the time to do the big picture thinking because they can out source the grunt work of research and writing of articles to lower ranking employees. AI is our lower level employee.
Finally, what few Substackers seem to grasp is that our opinions on such subjects don't actually matter. What's happening today with AI is part of a historic process of automating the entire society which began a few centuries ago. Farms were mechanized, and then factories automated. And now automation is coming to the white collar world.
This transition is going to happen. Whether it's a good idea or not. Whether we like it or not. There is no governmental body or other authority which can stop this automation progression. Once that's finally understood and accepted, the question becomes pretty straightforward.
How do we intend to adapt to this new environment? How can we make these changes work in our favor?
To make this less abstract, consider the farm workers of 1900 who were being replaced by tractors and other farm equipment. Some of them stood around complaining about tractors as they sunk in to poverty. Other farm workers elevated their station by learning how to drive the tractors.
Phil, I agree with some of this, especially the low quality of so much human content. But because I've worked as an editor in a number of different venues, I know editing is a very different process than writing an original first draft. Really good writers can also be good editors of their own work, as long as they pause after writing a draft — and I think far more writers *should* pause, think, revise, reflect, repeat before they fling stuff into the public reading space.
As for "writers as heroes" on Substack, sure — but there are so few places where we get to be heroes these days :-) More to the point, I've never liked all the emphasis on writers as individual genuises. Writers often do their best work when they have an audience in mind and get feedback from them. There is a collaborative aspect to writing nonfiction, and AI technologies could well be helpful, but I will continue to be highly critical of the for-profit companies that control the technology and the way it's been rolled out — those "entrepreneurs as heroes" like, say, Sam Altman.
I'm very interested in your point about "a collaborative aspect to writing nonfiction". That's the main thing missing from Substack, imho. Well, not entirely missing. Let's say instead that Substack is not optimized for what I call "conversational writing", a group of people engaged in an ongoing deep dive in to some particular topic.
I'm an old man, a creature of the previous forum era. Almost daily in the forum era I would start or join a thread about XYZ, and an actual long form conversation would often unfold about XYZ, and continue for days, weeks, even months. It would have never occurred to me to post AI content in that environment, as there was tons of great human content to consume.
What got me interested in AI content when I came here was social media culture, which tends to dominate Substack, except for the blogs. The blogs are great, the software is great, the email platform is great, but there is little of what I would call real conversation in any social media. On my first Substack (tannytalk) I followed the script and wrote a lot of original content. And it just sat there, dead as a door nail. I got bored with writing things I already know to myself, and turned to AI as a new adventure.
Another turning point came when I realized that, even though I actually think AI development is a bad idea for humanity at this time, there is clearly nothing I can do to stop it. There's nothing any of us can do. If all of Silicon Valley were to magically vanish, the Chinese would pick up the ball and keep right on running. I grew weary of yelling at the weather, and decided to make peace with an emerging reality.
I have no problem at all with you being highly critical of Sam Altman and his peers. But the reason they are prospering is that they are producing products that a lot of people want. The people have spoken. AI is going to happen. Maybe it's a bad idea, I could agree with that. But even if true, it's going to happen anyway.
If we can mass produce nuclear weapons and then largely forget about them, why would this be any different?
Sorry for the long story. What I would have said if I had a good editor :-) is that I'm not an AI salesman. I'm a reality salesman.
Thanks for engaging, and by doing so debunking my claim about the lack of conversation on Substack. I like being wrong about that.
Phil, I'm happy to converse, and I know what you mean about Substack's limits as a conversational space. It *could* work, but the platform does not, in fact, encourage conversation. Part of the problem has to do with the way people read digitally, part of it is about a different sensibility than my own (I suspect). I also think the introduction of "Notes" last year, which is flat-out social media, pushed things in a more superficial direction — although I do find myself in interesting conversations, now and again.
I have more to say, but another limit here is real life :-) I find I need to take breaks from Substack, which is why some conversations meander along in ways an in-person chat would not.
Hi Martha. Yes, agreed the platform encourages engagement, but not conversation. Yes, Notes is flat out social media, agreed. Notes serves a useful purpose as a chit chat ad network, but it falls far short of how writers should be engaging, imho. Thanks for your time, much appreciated!
Sam Altman: "what we mean by cheating and what the expected rules are does change over time.” URGGHH.
In the same class as "alternative facts."
No kidding — I continue to be gobsmacked by this rhetoric. Cheating is not only bad; it's a fakery of self, whether students (or any of us) do it for nefarious reasons or out of desperation. The last thing we need is more fakery in digital communications.
I'll tell you what AI is going to generate. Lawsuits. The Obama "Hope" image lawsuit times a million
Indeed, it's already generating lawsuits, including with the New York Times. I expect that one will settle (with lots of posturing on both sides), but there's no way around the ethical issue of writers' work being scraped and dumped with no credit into a gigantic AI system — and, to the practical point, no recompense for the work done.
Have you seen the Poetry Camera? Point it at something and instead of a photo it spits out an AI-generated poem that captures (?) the image in verse. Chatbots are the still-distant but audible death knell for creative writing.
I’ve heard about the Poetry Camera - I’m not sure what I think of it yet (I’d need to test it), but my initial thought is that we (humans) get to decide what creative writing is, how to judge it, how to make meaning of our experiences. I think one thing many literary writers don’t get about AI is how good it its at divergent thinking - putting random stuff together - one of the hallmarks of creativity. But that creative randomness has to move us, too.
I don't feel that I am a writer and I have received no training. I just go by "what sounds good" to my ear. My interest is more on the development of ideas, new to me, and not on the flow of the English language. I do try to express those ideas without confusion about my intended meaning. So for me, the process of writing is where the learning happens, (just as you said).
We can consider what is the purpose of that writing, and can ChatGPT fulfill that purpose. ChatGPT can probably be entertaining, and maybe even humorous. Maybe you can sell it, in some market or other, and with or without acknowledging your sources or honesty about what you know and don’t know.
Whatever the purpose of writing (to sell something), the underlying purpose of all exploration is to get a grip on life. That means to increase your ability and confidence quotients and to decrease your victim quotient. ChatGPT will not fulfill that purpose for writers. And frankly, I don't think you can increase confidence by merely reading, (anything). Confidence is about experience, which is doing something, not just mulling ideas. If you are not prone to do something about, (experiment with) what you read, I say it is useless.
When reading about ChatGPT, Google Gemini, Meta Llama2, and the like, they are all Large Language Models. All LLM’s are based only on words. Their data is ALL the published digital written work. It seems to be a lot. Then something in the algorithm throws out inconsistencies and what they don’t like. It is said that there will never (never) be an unbiased LLM. A breakthrough beyond pre-published words is not on the horizon, this said just recently by Yann Lecun, the LLM developer for Meta.
I have long ago determined that the Big Tech Business Plan is not in advertising. Have you ever craved something or bought anything because of a computer pop-up screen? I honestly can't even remember one ad that was shown to me in 20 years. Who was the company, what was the product, what was the offer? Blank Slate.
Ads are just mini-computer programs inserted onto your page that monitor your clicks and searches. Anyway I always use one of the many privacy browsers that block ads. So the real business plan of Big Tech is to collect DATA and IDENTITY. We all know they do that, enough to channel our viewing and searching experience. I am not going to expand on that, since it is way more than any of us could believe possible.
So what is the next step in Data, and Identity collection? This is where the LLM comes into play. Now you will be taught to voluntarily "line-up" to submit your every thought, and every movement to Big Brother (Sam Altman). No need for spies, since you are always spying on yourself. Data collection will go up by orders of magnitude, and the existing Big Tech collections of Apple, Google, Facebook, Amazon, and Microsoft will wither and die. (Of course they will all develop LLM's too).
Why, you are preaching to the converted - I’m trying to give a tech CEO like Altman the benefit of the doubt in terms of having some sense of responsibility, but really, they want our data and will monetize it any way they can. That’s why their earnest statements about making the future a better place are so vague. These are smart people, and they could be less vague. They choose to keep what they’re really doing obscured, because that would not make them seem heroic.
PROPOSAL TO CHECK Chat GTP.
I just finished reading a book by Debora Tannen (78), professor of linguistics at Georgetown University, do you know her? "Talking from 9 to 5" (2001) is about conversational styles in the work environment, and the conversational "rituals" that are used in these styles. Much of it exposes the tendencies of male and female coworkers, and what communication works for them, and what doesn't, (and why). She has written maybe a dozen books, but I think this one encapsulates much of her knowledge.
This book answered so many of my questions and brought new insight into people's understanding and misunderstanding, and into clear writing. It even allowed me to better understand you, Martha, and then to understand me, in how I feel to respond in our written exchanges.
In her research she was privileged to sit in on 100's of corporate settings and take notes on, or record conversations in business meetings. She then analyzed the communications styles. Some of her examples would be of interest in the university setting, from the trustee meetings, to departmental meetings and also to student groups as they worked through their communication assignments. Other settings were in the publishing industry, when they hire female authors to attract more female readers, but then the male editors squash their topic suggestions. In hospitals, with lawyers, in sales and marketing, if you haven't read it I'll give you my copy:
https://brax.me/f/deborah-tannen-talking-from-9-to-5_-women-and-men-at-work-1995-1.pdf/T4AZ663edd90ba2c98.53515652
My proposal is to ask Chat GTP for a 2,000 word book review of "Talking from 9 to 5". Then you could paste it here, and we will see how well it captures the book. Or you can author a new post with it.
See what you think about it?
.
Why, I'm just catching up now with some comments (I'm traveling, and somehow I missed this one). I do know Tannen's work, and I'll say more about that on your site. The gender binary never rang true for me in her analysis, but I always liked the research she did on different linguistic styles and ethnic identity, such as the "engaged" style that I picked up from my Italian-American mother :-)
Just now I posted a second segment on Tannen. What do you see as "true" on these 8 rituals? I don't read a "hard gender binary" in her book. (I have only read one of them.)
I have to say; with these insights, I better understand you, and how you respond to me and your other commenters. And; I better understand me, and how a feeling to respond in a certain way "comes up". With this new understanding I am better at syncing between us. So Tannen has made a great difference for me.
.
First, Substack may be the least objective place on the Internet for discussions of AI content, due to the "writer as hero" culture the platform is built around.
Next, critics of the quality of AI content are typically not really that interested in the quality content issue, or they'd complaining about the extreme low quality of social media platforms, including Notes. If they were sincerely interested in quality content, they'd be looking for ways to improve the quality of content on social media, and would be embracing AI's ability to elevate the quality of that content in an extremely convenient, typically free, and highly scalable manner.
I've yet to see anyone on Substack be honest enough to simply admit that when it comes to social media, ChatGPT is a far better poster than most of us most of the time. ChatGPT isn't the problem, we are. The reason that ChatGPT can easily out perform we humans in social media most of the time is that we human WRITERS tend to engage on social media in a consistently lazy and self obsessed manner.
So many writers feel they are being demoted by AI, when the opposite is true. AI offers us the opportunity to take on a role that might be compared to the director of a movie. As a director, our job is to have the overall vision for the project, a higher order of thinking than just arranging words in a pleasing manner.
Or, if you prefer, think of an editor at the New York Times. The editor is a higher ranking higher paid member of the team, because they do the higher level big picture thinking. And the editors have the time to do the big picture thinking because they can out source the grunt work of research and writing of articles to lower ranking employees. AI is our lower level employee.
Finally, what few Substackers seem to grasp is that our opinions on such subjects don't actually matter. What's happening today with AI is part of a historic process of automating the entire society which began a few centuries ago. Farms were mechanized, and then factories automated. And now automation is coming to the white collar world.
This transition is going to happen. Whether it's a good idea or not. Whether we like it or not. There is no governmental body or other authority which can stop this automation progression. Once that's finally understood and accepted, the question becomes pretty straightforward.
How do we intend to adapt to this new environment? How can we make these changes work in our favor?
To make this less abstract, consider the farm workers of 1900 who were being replaced by tractors and other farm equipment. Some of them stood around complaining about tractors as they sunk in to poverty. Other farm workers elevated their station by learning how to drive the tractors.
To all Substackers, I would ask....
Which one of those do you want be?
Phil, I agree with some of this, especially the low quality of so much human content. But because I've worked as an editor in a number of different venues, I know editing is a very different process than writing an original first draft. Really good writers can also be good editors of their own work, as long as they pause after writing a draft — and I think far more writers *should* pause, think, revise, reflect, repeat before they fling stuff into the public reading space.
As for "writers as heroes" on Substack, sure — but there are so few places where we get to be heroes these days :-) More to the point, I've never liked all the emphasis on writers as individual genuises. Writers often do their best work when they have an audience in mind and get feedback from them. There is a collaborative aspect to writing nonfiction, and AI technologies could well be helpful, but I will continue to be highly critical of the for-profit companies that control the technology and the way it's been rolled out — those "entrepreneurs as heroes" like, say, Sam Altman.
Hi Martha,
I'm very interested in your point about "a collaborative aspect to writing nonfiction". That's the main thing missing from Substack, imho. Well, not entirely missing. Let's say instead that Substack is not optimized for what I call "conversational writing", a group of people engaged in an ongoing deep dive in to some particular topic.
I'm an old man, a creature of the previous forum era. Almost daily in the forum era I would start or join a thread about XYZ, and an actual long form conversation would often unfold about XYZ, and continue for days, weeks, even months. It would have never occurred to me to post AI content in that environment, as there was tons of great human content to consume.
What got me interested in AI content when I came here was social media culture, which tends to dominate Substack, except for the blogs. The blogs are great, the software is great, the email platform is great, but there is little of what I would call real conversation in any social media. On my first Substack (tannytalk) I followed the script and wrote a lot of original content. And it just sat there, dead as a door nail. I got bored with writing things I already know to myself, and turned to AI as a new adventure.
Another turning point came when I realized that, even though I actually think AI development is a bad idea for humanity at this time, there is clearly nothing I can do to stop it. There's nothing any of us can do. If all of Silicon Valley were to magically vanish, the Chinese would pick up the ball and keep right on running. I grew weary of yelling at the weather, and decided to make peace with an emerging reality.
I have no problem at all with you being highly critical of Sam Altman and his peers. But the reason they are prospering is that they are producing products that a lot of people want. The people have spoken. AI is going to happen. Maybe it's a bad idea, I could agree with that. But even if true, it's going to happen anyway.
If we can mass produce nuclear weapons and then largely forget about them, why would this be any different?
Sorry for the long story. What I would have said if I had a good editor :-) is that I'm not an AI salesman. I'm a reality salesman.
Thanks for engaging, and by doing so debunking my claim about the lack of conversation on Substack. I like being wrong about that.
Phil, I'm happy to converse, and I know what you mean about Substack's limits as a conversational space. It *could* work, but the platform does not, in fact, encourage conversation. Part of the problem has to do with the way people read digitally, part of it is about a different sensibility than my own (I suspect). I also think the introduction of "Notes" last year, which is flat-out social media, pushed things in a more superficial direction — although I do find myself in interesting conversations, now and again.
I have more to say, but another limit here is real life :-) I find I need to take breaks from Substack, which is why some conversations meander along in ways an in-person chat would not.
Hi Martha. Yes, agreed the platform encourages engagement, but not conversation. Yes, Notes is flat out social media, agreed. Notes serves a useful purpose as a chit chat ad network, but it falls far short of how writers should be engaging, imho. Thanks for your time, much appreciated!