> Microsoft’s AI CEO is saying AI is going to take everybody’s job. And Sam Altman is saying that AI will wipe out entire categories of jobs. ANd Matt Shumer is saying that AI is currently like Covid in January 2020—as in, “kind of under the radar, but about to kill millions of people”.
> I legitimately feel like I am going insane when I hear AI technologists talk about the technology. They’re supposed to market it. But they’re instead saying that it is going to leave me a poor, jobless wretch, a member of the “permanent underclass,” as the meme on Twitter goes.
They are marketing it. The target customer isn't the user paying $20 for ChatGPT Pro, though; the customers are investors and CEOs, and their marketing is "AI is so powerful and destructive that if you don't invest in AI, you will be left behind." FOMO at its finest.
And a great deal of the talking shops/podcasts/keynotes/nonprofits around AI existential risk are all part of this same play. They receive funding from the large AI companies so that the latter can continue to talk their own books using this angle.
I hate LLMs because you can solve any problem that LLMs can solve in a much better way but people are too stupid, cheap or lazy to put in the effort to do so and share it with everyone.
That and the whitewashing it allows on layoffs from failing or poorly planned businesses.
> Being able to easily interact with banks, without waiting in a line that’s too long for the dum-dum you get at the end to be a real consolation, made people use banks more.
Actually, in my city, not the ATMs, but the apps which made it possible to do almost everything on the phone significantly reduced the number of banks in the last few years. I have to go very rarely to the bank, but, when I have to do, I see that another close one has closed and I have to go somewhere even farther.
The rule of metaprogramming is that it ends up just as convoluted and full of edge cases as regular code, just without a nice way to debug. The rule is also that it always seems like a fantastic idea at first and will solve so many isues.
I've been programming since 1994. I've seen a lot. I almost always end up despising any metaprogramming system and wish we'd just kept things simpler even if it meant boilerplate.
The AI executives are marketing it—it’s just none of us are the target demographic. They are marketing it to executives and financiers, the people who construct the machinations to keep their industry churning, and those who begrudge the necessity of labor in all its forms.
Sam Altman gave millions to Andrew Yang for pushign UBI, so they are trying to forewarn and experiment with finding the right solution. Most of the world prefers to shove their heads in the sand though and call them grifters, so of course we'll do nothing until it's catastrophic.
The amount of EM dashes and usage of negation does make me think AI wrote part of this. I'll give credit for lack of semicolons, but people are starting to get a bit better at "humanizing" their outputs.
The reason I dislike AI use in certain modes is because the end result looks like a Happy Meal toy from McDonalds. It looks roughly like the thing you wanted or expected, but on even a casual examination it falls far short. I don’t believe this is something we can overcome with better models. Or, if we can then what we will end up writing as prompts will begin to resemble a programming language. At which point it just isn’t worth what it costs.
This tech is a breakthrough for so many reasons. I’m just not worried about it replacing my job. Like, ever.
Whenever there is a massive paradigm shift in technology like we have with AI today, there are absolutely massive, devastating wars because the existing strategic stalemates are broken. Industrialized precision manufacturing? Now we have to figure out who can make the most rifles and machine guns. Industrialized manufacturing of high explosives? Time to have a whole world war about it. Industrialized manufacturing of electronics? Time for another world war.
Industrialized manufacturing of intelligence will certainly lead to a global scale conflict to see if anyone can win formerly unwinnable fights.
Thus the concerns about whether you have a job or not will, in hindsight, seem trivial as we transition to fighting for our very survival.
I have some friends who are embracing it and using it to transform their businesses (eg insurance sales), and others who hate it and think it should be banned (lawyers, white collar).
I think for a lot of people it feels like an inconvenient thing they have to contend with, and many are uncomfortable with rapid change.
The vibes around the self-driving car hype (maybe 10 years ago?) felt very similar to me, but on a smaller scale. There was a lot of "You might like driving your car and having a steering wheel, but if you do, you're a luddite who will soon be forced to ride about in our featureless rented robot pods" type of statements, or that one AI scientist who was quoted saying we should just change laws around how humans are allowed to interact with streets to protect the self-driving cars.
Not all of it was like that, I think oddly enough it was Tesla or just Elon Musk claimng you'd soon be able to take a nap in your car on your morning commute through some sort of Jetsons tube or that you could let your car earn money on the side while you weren't using it, which might actually be appealing to the average person. But a lot of it felt like self-driving car companies wanted you to feel like they just wanted to disrupt your life and take your things away.
> The classical cultural example is the Luddites, a social movement that failed so utterly
Maybe not the best example? The luddites were skilled weavers that had their livelihoods destroyed by automation. The govt deployed 12,000 troops against the luddites, executed dozens after show trials, and made machine breaking a capital offense.
Part of what's going on here -- why we have this gap in what we say we fear and how we act, is just a human deficiency.
I remember when Covid got out of control in China a lot of people around me [in NY] had this energy of "so what it'll never come to us." I'm not saying that they believed that, or had some rational opinion, but they had an emotional energy of "It's not big deal." The emotional response can be much slower than the intellectual response, even if that fuse is already lit and the eventuality is indisputable.
Some people are good at not having that disconnect. They see the internet in 1980 and they know that someday 60 years from now it'll be the majority of shopping, even though 95% of people they talk to don't know what it is and laugh about it.
AI is a little-bit in that stage... It's true that most people know what it is, but our emotional response has not caught up to the reality of all of the implications of thinking machines that are gaining 5+ iq points per year.
> And Matt Shumer is saying that AI is currently like Covid in January 2020—as in, "kind of under the radar, but about to kill millions of people"
This is where the misrepresentation... no, the lie comes in. It always does in these "sensible middle" posts! the genre requires flattening both sides into dumber versions of themselves to keep the author positioned between two caricatures. Supremely done, OP.
If you read Matt's original article[0] you see he was saying something very different. Not "AI is going to kill lots of people" but that we're at the point on an exponential curve where correct modeling looks indistinguishable from paranoia to anyone reasoning from base rates of normal experience. The analogy is about the epistemic position of observers, not about body counts.
The cracks are showing, and all the “AI is going to eliminate 50% of white collar jobs” fear mongering is simply signaling we’re in the final stages before the bubble implosion.
The AI bros desperately need everyone to believe this is the future. But the data just isn’t there to support it. More and more companies are coming out saying AI was good to have, but the mass productivity gains just aren’t there.
A bunch of companies used AI as an excuse to do mass layoffs only to then have to admit this was basically just standard restructuring and house cleaning (eg Amazon).
Theres so much focus on white collar jobs in the US but these have already been automated and offshored to death. What’s there now is truly a survival of the fittest. Anything that’s highly predicable, routine, and fits recurring patterns (ie what AI is actually good at) was long since offshored to places like India. To the extent that AI does cause mass disruption to jobs, the India tech and BPO sectors would be ground zero… not white collar jobs in the US.
The AI bros are in a fight for their careers and the signal is increasingly pointing to the most vulnerable roles out there at the moment being all those tangentially tacked onto the AI hype cycle. If real measurable value doesn’t show up very soon (likely before year end) the whole party will come crashing down hard.
Right now is the good time for the job market. The S&P is at an all time high.
In the next recession, I expect massive layoffs in white collar work and there is no way those jobs are coming back on the other side.
40-50% of US white collar work hours are spent on procedural, rules-based tasks. Then another large chunk is managing the people doing procedural, rules-based tasks and support of people doing rules based tasks. Salary and benefits are 50% of operating costs for most business.
Maybe you do something really interesting and unique but that is just not what most white collar workers in the US are doing.
I know for myself, these are the final days of white collar work before I am unemployable as a white collar worker. I don't think the company I work for will exist either in 5 years. It is not a matter of Claude code being able to update a legacy system or not. It is that the tide hasn't really gone out in 15 years and all these zombie companies are going to get wiped out at the same time AI is automating the white collar jobs. Delaying the business cycle from clearing over and over is not a free lunch, it is a bill that has been stacking up for a long time.
On the other side, the business as usual of today won't be an option.
From my own white collar experience, I think if you view procedural rules-based tasks as a graph, the automation of any one task depends so much on other tasks being automated. So it will seem like the automation is not working but at some point you get a contagion of automation. Then so much automation will happen at once.
It's very useful as a coding autocomplete. It provides a fast way to connect multiple disparate search criteria in one query.
It also has caused massive price hikes for computer components, negatively impacted the environment, and most importantly, subtly destroys people's ability to understand.
I've come to the same conclusions. I don't think the feelings are complicated though. It's just personal when you find use in it. It's kind of like cryptocurrency in the way that it doesn't democratize efficiency and usefulness amongst everyone. So for a few it is a powerful and useful tool, but ultimately at the cost of everything else.
Once I read a reference to Clavicular, I realized that the very first thing this author should do is stop reading the NYTimes. If the goal is to experience things closer to reality haha.
> I have a friend who is a new TA at a university in California. They’ve had to report several students, every semester, for basically pasting their assignments into ChatGPT.
We've solved this problem before.
You have 2 separate segments:
1. Lessons that forbid AI
2. Lessons that embrace AI
This doesn't seem that difficult to solve. You handle it like how you handle calculators and digital dictionaries in universities.
Moving forward, people who know fundamentals and AI will be more productive. The universities should just teach both.
55 comments
[ 4.2 ms ] story [ 69.1 ms ] thread> I legitimately feel like I am going insane when I hear AI technologists talk about the technology. They’re supposed to market it. But they’re instead saying that it is going to leave me a poor, jobless wretch, a member of the “permanent underclass,” as the meme on Twitter goes.
They are marketing it. The target customer isn't the user paying $20 for ChatGPT Pro, though; the customers are investors and CEOs, and their marketing is "AI is so powerful and destructive that if you don't invest in AI, you will be left behind." FOMO at its finest.
That and the whitewashing it allows on layoffs from failing or poorly planned businesses.
Human issues as always.
Actually, in my city, not the ATMs, but the apps which made it possible to do almost everything on the phone significantly reduced the number of banks in the last few years. I have to go very rarely to the bank, but, when I have to do, I see that another close one has closed and I have to go somewhere even farther.
Poor author, never tried expressive high-level languages with metaprogramming facilities that do not result in boring and repetitive boilerplate.
I've been programming since 1994. I've seen a lot. I almost always end up despising any metaprogramming system and wish we'd just kept things simpler even if it meant boilerplate.
This tech is a breakthrough for so many reasons. I’m just not worried about it replacing my job. Like, ever.
Hahah, this guy Gen-Zs.
Whenever there is a massive paradigm shift in technology like we have with AI today, there are absolutely massive, devastating wars because the existing strategic stalemates are broken. Industrialized precision manufacturing? Now we have to figure out who can make the most rifles and machine guns. Industrialized manufacturing of high explosives? Time to have a whole world war about it. Industrialized manufacturing of electronics? Time for another world war.
Industrialized manufacturing of intelligence will certainly lead to a global scale conflict to see if anyone can win formerly unwinnable fights.
Thus the concerns about whether you have a job or not will, in hindsight, seem trivial as we transition to fighting for our very survival.
I think for a lot of people it feels like an inconvenient thing they have to contend with, and many are uncomfortable with rapid change.
Not all of it was like that, I think oddly enough it was Tesla or just Elon Musk claimng you'd soon be able to take a nap in your car on your morning commute through some sort of Jetsons tube or that you could let your car earn money on the side while you weren't using it, which might actually be appealing to the average person. But a lot of it felt like self-driving car companies wanted you to feel like they just wanted to disrupt your life and take your things away.
Maybe not the best example? The luddites were skilled weavers that had their livelihoods destroyed by automation. The govt deployed 12,000 troops against the luddites, executed dozens after show trials, and made machine breaking a capital offense.
Is that what you have planned for me?
I remember when Covid got out of control in China a lot of people around me [in NY] had this energy of "so what it'll never come to us." I'm not saying that they believed that, or had some rational opinion, but they had an emotional energy of "It's not big deal." The emotional response can be much slower than the intellectual response, even if that fuse is already lit and the eventuality is indisputable.
Some people are good at not having that disconnect. They see the internet in 1980 and they know that someday 60 years from now it'll be the majority of shopping, even though 95% of people they talk to don't know what it is and laugh about it.
AI is a little-bit in that stage... It's true that most people know what it is, but our emotional response has not caught up to the reality of all of the implications of thinking machines that are gaining 5+ iq points per year.
We should be starting to write the laws now.
This is where the misrepresentation... no, the lie comes in. It always does in these "sensible middle" posts! the genre requires flattening both sides into dumber versions of themselves to keep the author positioned between two caricatures. Supremely done, OP.
If you read Matt's original article[0] you see he was saying something very different. Not "AI is going to kill lots of people" but that we're at the point on an exponential curve where correct modeling looks indistinguishable from paranoia to anyone reasoning from base rates of normal experience. The analogy is about the epistemic position of observers, not about body counts.
[0]: https://shumer.dev/something-big-is-happening
The AI bros desperately need everyone to believe this is the future. But the data just isn’t there to support it. More and more companies are coming out saying AI was good to have, but the mass productivity gains just aren’t there.
A bunch of companies used AI as an excuse to do mass layoffs only to then have to admit this was basically just standard restructuring and house cleaning (eg Amazon).
Theres so much focus on white collar jobs in the US but these have already been automated and offshored to death. What’s there now is truly a survival of the fittest. Anything that’s highly predicable, routine, and fits recurring patterns (ie what AI is actually good at) was long since offshored to places like India. To the extent that AI does cause mass disruption to jobs, the India tech and BPO sectors would be ground zero… not white collar jobs in the US.
The AI bros are in a fight for their careers and the signal is increasingly pointing to the most vulnerable roles out there at the moment being all those tangentially tacked onto the AI hype cycle. If real measurable value doesn’t show up very soon (likely before year end) the whole party will come crashing down hard.
Right now is the good time for the job market. The S&P is at an all time high.
In the next recession, I expect massive layoffs in white collar work and there is no way those jobs are coming back on the other side.
40-50% of US white collar work hours are spent on procedural, rules-based tasks. Then another large chunk is managing the people doing procedural, rules-based tasks and support of people doing rules based tasks. Salary and benefits are 50% of operating costs for most business.
Maybe you do something really interesting and unique but that is just not what most white collar workers in the US are doing.
I know for myself, these are the final days of white collar work before I am unemployable as a white collar worker. I don't think the company I work for will exist either in 5 years. It is not a matter of Claude code being able to update a legacy system or not. It is that the tide hasn't really gone out in 15 years and all these zombie companies are going to get wiped out at the same time AI is automating the white collar jobs. Delaying the business cycle from clearing over and over is not a free lunch, it is a bill that has been stacking up for a long time.
On the other side, the business as usual of today won't be an option.
From my own white collar experience, I think if you view procedural rules-based tasks as a graph, the automation of any one task depends so much on other tasks being automated. So it will seem like the automation is not working but at some point you get a contagion of automation. Then so much automation will happen at once.
It's very useful as a coding autocomplete. It provides a fast way to connect multiple disparate search criteria in one query.
It also has caused massive price hikes for computer components, negatively impacted the environment, and most importantly, subtly destroys people's ability to understand.
We've solved this problem before.
You have 2 separate segments:
1. Lessons that forbid AI 2. Lessons that embrace AI
This doesn't seem that difficult to solve. You handle it like how you handle calculators and digital dictionaries in universities.
Moving forward, people who know fundamentals and AI will be more productive. The universities should just teach both.