I almost feel like the NYT article (and yours to a lesser extent) is missing the forest for the trees. Its an obvious journalistic attempt to draw fear from people. And I think if it was framed better, an article like this could get my attention. I think the general discourse should be around how these LLMs affect human social health. These deaths are horrible, and to your point, should not be seen as quick statistics against or for LLMs. But a discourse around how people interact with these technologies and how they shape their beliefs, social happiness and more should be studied more fervently. I think making an article for a few deaths really undermines how people, particularly children, will develop emotional relationships with these models that aren't exactly known how they work.
It's been the case for while that MSM is drowning so they have to continue to push narratives that drive clicks and create revenue. One of the reasons I am enjoying Substack is clearly the freedom to write what you want with no editorial constraints and get instant feedback as to whether or not your point of view is valid or at least can find a readership. I do think there is a potential story in there, but I agree this isn't it - the goal clearly was to incentivize the pile-on towards the negatives of AI (of which there are many) but cherry picking a few examples of people who were likely suffering from mental health crises prior to using AI is misleading to say the least. I do like the chart - I teach a class where we often discuss media bias and I will add this to my arsenal of similar material demonstrating how much what we consume significantly distorts our perception of reality. I assume your familiar with Factfulness which makes similar points. Nice post.
I will typically always defend most writers/ journalists on these kinds of points.
They are not the problem, we are!
They write about whatever we read. We know that it's this way and not the other way because on *all* news sites there *are* articles that are 'sensible' and 'contextual' etc: people just don't get read.
I'm involved in the Nutrition world (mid-MSc), and this is arguably even more prominent here. If you let it it will drive you crazy and you'll become incredibly frustrated (arguably rightly so), but at the end of the day it is what it is: biology
I also feel that 'misleading articles' are actually *good* for writers who *do* have domain knowledge and *can* help contextualise etc: they are providing more material for you to write and expand on!
It's just basic biology/evolution: for us novelty is interesting, and for our ancestors it was risky and thus we evolved to pay extra attention.
Again, 2 things:
1. The 'proper' articles do exist, they just don't get read thus no incentive to write more
2. Writers with greater domain-knowledge (which we cannot necessarily expect the journalist to know) benefit more than anyone from said 'poor' articles! Nutrition is AI x100!
I guess we disagree. 1) The NYT doesn't need clickbait. 2) attention-seeking is not omnipotent; there are great articles that are also popular that are not clickbait. Clickbait is merely the easy route. Taken by those who have little respect for truth (or hidden interests)
Disagreements are good! As said, that’s why we’re paying to be here on your Substack allowing you to make a very comfortable income!
Writers, especially employed ones, write about what humans read. We can’t expect journalists to want to write about “another million people die from CVD that was entirely preventable”.
But it's funny that you talk about "employed writers" as if we're talking here about a small outlet. This is the New York Times. NYT journalists have more freedom than pretty much *any other news writer on this planet* to not go clickbait -- surely more than me!
Perhaps! Although I could see it being the other way around as well: write about what gets hits or you risk getting demoted with 1,000 other journalists eyeing up your position.
It took me a while to get past those charts. Even more stark than the one featured in this piece was one in the Helen Ritchey article showing global causes of death. Noncommunicable disease is the big baddie with 74% of attributed deaths. 0.2% are attributed to war (as of 2019).
Just about every night I hear from my wife about how the world is going to hell. She is probably a lot worse than most at resonating with bad news in the media. Fear and stress from this kind of media spin is probably having a much bigger impact on her likely lifespan than any of the things she fears.
The larger point of your essay is excellent, however, and I will use this when I encounter the scare stories of how chatbots are the greatest nemesis ever known to mankind. They may become so, but they have a long way to go to dethrone the champs.
Exactly. The broader point is much more important than anything AI related actually. I love our world in data and similar websites because they illuminate aspects of reality no one else can. They do gods work
Does anything merely become popular anymore? They seem to only become “viral”.
Is it just performative linguistic laziness, much in the way albums and movies aren't released anymore, they are merely "dropped"?
But 'viral' has real meaning. Have all we forgotten what an R0 represents?
Mental health crises might be more common among younger people than before, but that doesn’t make them “viral”.
If there isn’t network-effect growth, “viral” has little to do with it. And word-of-mouth or a review in the New York Times is not a measure of "viral" anymore than the 80s movie The Princess Bride was "viral".
I almost feel like the NYT article (and yours to a lesser extent) is missing the forest for the trees. Its an obvious journalistic attempt to draw fear from people. And I think if it was framed better, an article like this could get my attention. I think the general discourse should be around how these LLMs affect human social health. These deaths are horrible, and to your point, should not be seen as quick statistics against or for LLMs. But a discourse around how people interact with these technologies and how they shape their beliefs, social happiness and more should be studied more fervently. I think making an article for a few deaths really undermines how people, particularly children, will develop emotional relationships with these models that aren't exactly known how they work.
I agree. But my article is not really about that. I'm merely calling out legacy media for being so sensationalistic. Anyway, to your last sentence: 💯
It's been the case for while that MSM is drowning so they have to continue to push narratives that drive clicks and create revenue. One of the reasons I am enjoying Substack is clearly the freedom to write what you want with no editorial constraints and get instant feedback as to whether or not your point of view is valid or at least can find a readership. I do think there is a potential story in there, but I agree this isn't it - the goal clearly was to incentivize the pile-on towards the negatives of AI (of which there are many) but cherry picking a few examples of people who were likely suffering from mental health crises prior to using AI is misleading to say the least. I do like the chart - I teach a class where we often discuss media bias and I will add this to my arsenal of similar material demonstrating how much what we consume significantly distorts our perception of reality. I assume your familiar with Factfulness which makes similar points. Nice post.
Thanks Stephen! I'm not familiar but I've read Steven Pinker's Enlightenment Now which makes similar arguments
"This is what drives views"
I will typically always defend most writers/ journalists on these kinds of points.
They are not the problem, we are!
They write about whatever we read. We know that it's this way and not the other way because on *all* news sites there *are* articles that are 'sensible' and 'contextual' etc: people just don't get read.
I'm involved in the Nutrition world (mid-MSc), and this is arguably even more prominent here. If you let it it will drive you crazy and you'll become incredibly frustrated (arguably rightly so), but at the end of the day it is what it is: biology
I also feel that 'misleading articles' are actually *good* for writers who *do* have domain knowledge and *can* help contextualise etc: they are providing more material for you to write and expand on!
But what do you think about Our World in Data's chart? Can you defend the NYT on that because "we" somehow are the problem? Not even Google searches!!
Of course!
It's just basic biology/evolution: for us novelty is interesting, and for our ancestors it was risky and thus we evolved to pay extra attention.
Again, 2 things:
1. The 'proper' articles do exist, they just don't get read thus no incentive to write more
2. Writers with greater domain-knowledge (which we cannot necessarily expect the journalist to know) benefit more than anyone from said 'poor' articles! Nutrition is AI x100!
I guess we disagree. 1) The NYT doesn't need clickbait. 2) attention-seeking is not omnipotent; there are great articles that are also popular that are not clickbait. Clickbait is merely the easy route. Taken by those who have little respect for truth (or hidden interests)
Disagreements are good! As said, that’s why we’re paying to be here on your Substack allowing you to make a very comfortable income!
Writers, especially employed ones, write about what humans read. We can’t expect journalists to want to write about “another million people die from CVD that was entirely preventable”.
And I thank you for your support!!
But it's funny that you talk about "employed writers" as if we're talking here about a small outlet. This is the New York Times. NYT journalists have more freedom than pretty much *any other news writer on this planet* to not go clickbait -- surely more than me!
Perhaps! Although I could see it being the other way around as well: write about what gets hits or you risk getting demoted with 1,000 other journalists eyeing up your position.
It took me a while to get past those charts. Even more stark than the one featured in this piece was one in the Helen Ritchey article showing global causes of death. Noncommunicable disease is the big baddie with 74% of attributed deaths. 0.2% are attributed to war (as of 2019).
Just about every night I hear from my wife about how the world is going to hell. She is probably a lot worse than most at resonating with bad news in the media. Fear and stress from this kind of media spin is probably having a much bigger impact on her likely lifespan than any of the things she fears.
The larger point of your essay is excellent, however, and I will use this when I encounter the scare stories of how chatbots are the greatest nemesis ever known to mankind. They may become so, but they have a long way to go to dethrone the champs.
Exactly. The broader point is much more important than anything AI related actually. I love our world in data and similar websites because they illuminate aspects of reality no one else can. They do gods work
Does anything merely become popular anymore? They seem to only become “viral”.
Is it just performative linguistic laziness, much in the way albums and movies aren't released anymore, they are merely "dropped"?
But 'viral' has real meaning. Have all we forgotten what an R0 represents?
Mental health crises might be more common among younger people than before, but that doesn’t make them “viral”.
If there isn’t network-effect growth, “viral” has little to do with it. And word-of-mouth or a review in the New York Times is not a measure of "viral" anymore than the 80s movie The Princess Bride was "viral".
Alberto, Kent Taylor here. Father of Alex.
Reach out to me if you want to discuss.