"now is the time to stand athwart our slide into dystopia, yelling “Stop!" Who would have guessed that the reality of modern life has so mugged our belovedly progressive Mari that she has adopted the stance of 1955 William F Buckley. Turn, turn, turn indeed. Welcome to my world of old curmudgeons who remember a better time . . . and long for it again.
I’m so happy you caught the reference! (I originally had a link to a citation, but then I worried that that would be patronizing, so I took it out. You have vindicated me!) I am as liberal as they come, but even I know that not all change is good, especially when it comes to those horrible Apple goggles!
The thing that I'm uncomfortable with when I see these goggles, and Google glass before, is that I may be being recorded at any point. Google glass in particular had this creepy "I may or may not be recording you in the public bathrooms" vibe about it.
I find myself of mixed opinions about that poem. On one hand, yes, you do want people to show up. On the other hand, all having to write essays in school did was give me a HATRED of writing and a STEMlord's conviction that English is for pretentious wankers. I would have happily told AI "give me 500 words on Biblical symbolism in Hugo" and gotten some sleep, probably gotten out ahead.
I'm right in the target demographics for early AI adopters, yet I feel myself hesitant. On one hand, I'm a hikkikomori with few friends (and none IRL), and as far as I can tell, most of the interactions people have with each other are fake, anyway. Yeah, LLMs are the definition of "confidently wrong," but people are full of shit too. Yeah, you'll never make a genuine connection with a machine...wait, I've never made a genuine connection with a human being. The only reason I'm not climbing down that rabbit hole is because there isn't a single client I'd trust to remain open source, transparent, and NOT for profit.
I am so sorry and sad to hear about your lack of friends and connections with other people. I have no advice, but just sympathy. I think the way we have set up our society in recent decades makes it much more difficult to make friends.
As for the essays, even though I’m a former English teacher (high school and college), I agree with you. I think we teachers rely too much on the five-paragraph essay format and also don’t do enough to communicate the value of the format for persuading readers. English and history teachers will have to adapt to AI and come up with assignments that challenge students’ creativity. And that will be a good thing.
I don't think that's a true "underpants gnome" scenario, because the ??? part can easily be conceived of. Your objection is that, at the moment, the quality of the work in Step 2 is poor. But increasing the quality of it is not too difficult, so if that Step 2 is the weak spot then you have very little bulwark against Step 3. More recent models hallucinate much less frequently, so this critique will likely be outdated by the same time next year.
I could easily spot the AI paragraph, but once again you need to consider that those capabilities also are improving, and if you took a newer model AI and trained it with substantial amounts of your writing this would be less obvious. The thing that will hold consumer-level AI's back from copying idiosyncratic author voices will probably be the limiting specifications packaged into them for public consumption.
Your general perspective here is good, but I don't think you can count on AI capabilities to remain this bad, you will have to fight the fight against increasingly powerful and useful models tempting people to rely upon them and move further apart out of convenience, or demanding they adopt it to keep their jobs for just a little bit longer before being replaced.
I think your last paragraph is the key, and we’re already seeing it happening in the workplace, where many jobs (especially in coding) expect that workers will use AI. My underpants gnomes analogy was for the paperclip-maximizer scenario, in which the AIs become sentient and kill all life on earth. I am by nature a hopeful person, so this scenario just seems fantastical to me. But the rest of what you describe? Definitely, it is coming for us.
yes lets say no to dystopia, the future is looking bleak and sanitised with nothing left to chance. As an artist I want to continue to challenge myself, to learn to struggle with creation. I don’t want a bloody machine to do everything for me. Not sure the younger generation understands what they will miss relying on AI but we can teach them.
In my own attempt to have my regular life be less dependent on the internet, I have returned to a paper calendar for 2025. It's a bit of an adjustment, but I like it already because I find I remember my schedule better and I remember better what I write in it. I find the intrusion of AI into healthcare as a way to replace people very depressing. I don't know why the people in Silicon Valley are in such a rush to have machines take over jobs that require and depend on human connection. Bleahhh. Thanks for this column, Mari!
This is fascinating, because I have gone back to a paper calendar too! I find it easier to look at a calendar on my kitchen table than to have to open up an app on my phone (and I found all those pop-up reminders on my phone to be annoying). And I totally agree about AI in healthcare. Human connection is a necessary part of the cure!
Ooooh! Great minds think alike...Sorry for my slowness. The only downside of trying to be less online is that I sometimes don't see wonderful messages from friends--like this one!
I'm so pleased you referenced the Underpants Gnomes! My friend and I have a running joke about Step 2 and what it might be for whatever the project of the moment is. That was a delight 😊
To address the dystopian slide- I think the vibe of the coming time will be "We weren't made to live like this." People seem to be becoming more aware of that...
AI may not be *infinitely* patient yet. There was a case where some student asked Google's AI to do his homework for him, question by question, and at some point the AI told him he was a stain on the universe and he should kill himself. Google has confirmed this was real and that, in a very small number of cases, AI can go off the rails. Maybe schools should think about this before offering their own school-based AI personal teacher.
I guessed correctly on the AI paragraph. It was not only out of phase with your writing (which I'm well-acquainted with); it also had that bland and soulless flattened-aerial-view-of-the-three-dimensional-living-human-landscape "voice".
I see this in student work sometimes. It's unmistakable...so far.
I wish that students who were tempted to use AI to write their papers for them would look at a side-by-side comparison of writing by humans and writing by AI. It is so obvious that the AI writing is, as you say, "bland and soulless."
Great essay. As I read through the chatbot paragraph I was thinking that the tone seemed out of place for you. When you reported it later I didn't even have to look back, and realized, ah ha! Very funny.
When you said there was an AI-generated paragraph, I went back and picked it out quickly with a high degree of confidence (I was right! Yay me!), but just reading through without looking for it I wouldn't have noticed to be honest.
At my last physical the doctor asked if she could record our discussion. The session had a few extended side conversation. At the end of the session she showed me how AI had captured our discussion, edited out what wasn't impt. and summarized the medical aspects which she put in the chart. She said this was saving her a great deal of time.
This happened to me, but the visit summary included a comment that was obvious kidding-around banter about anxiety and an off-hand comment about chocolate covered almonds. The technology still needs human eyes/ears for something as important as medical records that influence future life/death decisions.
There are definitely ways that AI can be incredibly useful, like in this case. Your doctor rightly focused on the human interaction and let the AI do the secretarial work so she could pay attention to you. That is the right way for doctors to use AI. My dad’s dermatologist got it exactly backwards in my opinion when they use AI to interact (badly) with patients.
"now is the time to stand athwart our slide into dystopia, yelling “Stop!" Who would have guessed that the reality of modern life has so mugged our belovedly progressive Mari that she has adopted the stance of 1955 William F Buckley. Turn, turn, turn indeed. Welcome to my world of old curmudgeons who remember a better time . . . and long for it again.
I’m so happy you caught the reference! (I originally had a link to a citation, but then I worried that that would be patronizing, so I took it out. You have vindicated me!) I am as liberal as they come, but even I know that not all change is good, especially when it comes to those horrible Apple goggles!
The thing that I'm uncomfortable with when I see these goggles, and Google glass before, is that I may be being recorded at any point. Google glass in particular had this creepy "I may or may not be recording you in the public bathrooms" vibe about it.
Oh yeah, that is an excellent point.
I find myself of mixed opinions about that poem. On one hand, yes, you do want people to show up. On the other hand, all having to write essays in school did was give me a HATRED of writing and a STEMlord's conviction that English is for pretentious wankers. I would have happily told AI "give me 500 words on Biblical symbolism in Hugo" and gotten some sleep, probably gotten out ahead.
I'm right in the target demographics for early AI adopters, yet I feel myself hesitant. On one hand, I'm a hikkikomori with few friends (and none IRL), and as far as I can tell, most of the interactions people have with each other are fake, anyway. Yeah, LLMs are the definition of "confidently wrong," but people are full of shit too. Yeah, you'll never make a genuine connection with a machine...wait, I've never made a genuine connection with a human being. The only reason I'm not climbing down that rabbit hole is because there isn't a single client I'd trust to remain open source, transparent, and NOT for profit.
I am so sorry and sad to hear about your lack of friends and connections with other people. I have no advice, but just sympathy. I think the way we have set up our society in recent decades makes it much more difficult to make friends.
As for the essays, even though I’m a former English teacher (high school and college), I agree with you. I think we teachers rely too much on the five-paragraph essay format and also don’t do enough to communicate the value of the format for persuading readers. English and history teachers will have to adapt to AI and come up with assignments that challenge students’ creativity. And that will be a good thing.
I don't think that's a true "underpants gnome" scenario, because the ??? part can easily be conceived of. Your objection is that, at the moment, the quality of the work in Step 2 is poor. But increasing the quality of it is not too difficult, so if that Step 2 is the weak spot then you have very little bulwark against Step 3. More recent models hallucinate much less frequently, so this critique will likely be outdated by the same time next year.
I could easily spot the AI paragraph, but once again you need to consider that those capabilities also are improving, and if you took a newer model AI and trained it with substantial amounts of your writing this would be less obvious. The thing that will hold consumer-level AI's back from copying idiosyncratic author voices will probably be the limiting specifications packaged into them for public consumption.
Your general perspective here is good, but I don't think you can count on AI capabilities to remain this bad, you will have to fight the fight against increasingly powerful and useful models tempting people to rely upon them and move further apart out of convenience, or demanding they adopt it to keep their jobs for just a little bit longer before being replaced.
I think your last paragraph is the key, and we’re already seeing it happening in the workplace, where many jobs (especially in coding) expect that workers will use AI. My underpants gnomes analogy was for the paperclip-maximizer scenario, in which the AIs become sentient and kill all life on earth. I am by nature a hopeful person, so this scenario just seems fantastical to me. But the rest of what you describe? Definitely, it is coming for us.
yes lets say no to dystopia, the future is looking bleak and sanitised with nothing left to chance. As an artist I want to continue to challenge myself, to learn to struggle with creation. I don’t want a bloody machine to do everything for me. Not sure the younger generation understands what they will miss relying on AI but we can teach them.
I agree. The struggle is the reason we feel pride in our achievements! Achievements that come easily feel hollow.
In my own attempt to have my regular life be less dependent on the internet, I have returned to a paper calendar for 2025. It's a bit of an adjustment, but I like it already because I find I remember my schedule better and I remember better what I write in it. I find the intrusion of AI into healthcare as a way to replace people very depressing. I don't know why the people in Silicon Valley are in such a rush to have machines take over jobs that require and depend on human connection. Bleahhh. Thanks for this column, Mari!
This is fascinating, because I have gone back to a paper calendar too! I find it easier to look at a calendar on my kitchen table than to have to open up an app on my phone (and I found all those pop-up reminders on my phone to be annoying). And I totally agree about AI in healthcare. Human connection is a necessary part of the cure!
Ooooh! Great minds think alike...Sorry for my slowness. The only downside of trying to be less online is that I sometimes don't see wonderful messages from friends--like this one!
I'm so pleased you referenced the Underpants Gnomes! My friend and I have a running joke about Step 2 and what it might be for whatever the project of the moment is. That was a delight 😊
To address the dystopian slide- I think the vibe of the coming time will be "We weren't made to live like this." People seem to be becoming more aware of that...
Good point. Most of us are starting to feel uneasy about the profoundly unnatural, uncanny-valley aspect of a lot of this online life.
AI may not be *infinitely* patient yet. There was a case where some student asked Google's AI to do his homework for him, question by question, and at some point the AI told him he was a stain on the universe and he should kill himself. Google has confirmed this was real and that, in a very small number of cases, AI can go off the rails. Maybe schools should think about this before offering their own school-based AI personal teacher.
Oh wow, what a terrible story. Yeah, schools ought to take this incident as a cautionary tale!
I guessed correctly on the AI paragraph. It was not only out of phase with your writing (which I'm well-acquainted with); it also had that bland and soulless flattened-aerial-view-of-the-three-dimensional-living-human-landscape "voice".
I see this in student work sometimes. It's unmistakable...so far.
I wish that students who were tempted to use AI to write their papers for them would look at a side-by-side comparison of writing by humans and writing by AI. It is so obvious that the AI writing is, as you say, "bland and soulless."
Great essay. As I read through the chatbot paragraph I was thinking that the tone seemed out of place for you. When you reported it later I didn't even have to look back, and realized, ah ha! Very funny.
Glad you spotted it! So far everyone has been able to find the AI paragraph right away.
When you said there was an AI-generated paragraph, I went back and picked it out quickly with a high degree of confidence (I was right! Yay me!), but just reading through without looking for it I wouldn't have noticed to be honest.
Oh interesting! I suspect that if we aren’t specifically looking to identify AI-generated content, we might well not notice it.
At my last physical the doctor asked if she could record our discussion. The session had a few extended side conversation. At the end of the session she showed me how AI had captured our discussion, edited out what wasn't impt. and summarized the medical aspects which she put in the chart. She said this was saving her a great deal of time.
This happened to me, but the visit summary included a comment that was obvious kidding-around banter about anxiety and an off-hand comment about chocolate covered almonds. The technology still needs human eyes/ears for something as important as medical records that influence future life/death decisions.
Exactly! Wisdom and judgment are important—and so far beyond the capabilities of AI.
There are definitely ways that AI can be incredibly useful, like in this case. Your doctor rightly focused on the human interaction and let the AI do the secretarial work so she could pay attention to you. That is the right way for doctors to use AI. My dad’s dermatologist got it exactly backwards in my opinion when they use AI to interact (badly) with patients.