I'm sympathetic to the author's core position. I think LLMs as they're currently deployed are doing net societal harm so far. For example, for every React app whistled up in five hours, there's a high schooler or college student out there sleepwalking through their foundational courses and being set up to catastrophically fail upon trying to enter the job market.
But I think it's important for the quality of public discourse that we forward only the best, strongest arguments for any given position (whether I personally agree with that position or not). A weak argument gives unnecessary purchase to unscrupulous haters who might use that weaker argument as a low-grade "straw man" to dismiss the whole position.
So with that preamble, some constructive criticism (very well-intentioned):
> I am not exaggerating: We need to resist “AI” so that thousands, perhaps millions of people do not die.
On this one you are, in fact, exaggerating (certainly with "millions," likely with "thousands" as well).
The overwhelming majority of people who lose their job for whatever reason (and there are many reasons, not just this one issue) fall back on a support network of their family and friends while they "get back on their feet" by pivoting to another job or form of employment (sometimes branching into entrepreneurship, for example).
The alternative to employment at whatever FAANG or FAANG-adjacent company someone was laid off from is not "death." It is typically "some, sometimes quite significant, discomfort." Granted "AI Layoffs" are starting to ripple out beyond FAANG, but the story is similar in other industries.
If "layoffs kill people" is the moral hill we choose to plant our flag on, then mustn't we also campaign with equal vociferousness against all the other causes of layoffs in general? Since to do otherwise is to sit by while those laid off allegedly die in droves of thousands or perhaps millions?
It also occurs to me that this framing ("AI causes layoffs") also gives incompetent and unaccountable executives a pass by pinning the actual harm done by irresponsible layoffs (which to be clear, is not none!) on "the AI" rather than on the leaders who are in the positions to actually decide what happens in their organizations.
I think the argument is much stronger if we talk about the aggregate cost to humanity being paid in terms of displacement, life disruption, and curtailed opportunity (especially for juniors). We should not make the "toxic escalationism" move of equivocating things we don't like with negligent homicide, even if it makes for a punchy one-liner.
> For example, for every React app whistled up in five hours, there's a high schooler or college student out there sleepwalking through their foundational courses and being set up to catastrophically fail upon trying to enter the job market.
> forward only the best, strongest arguments for any given position... A weak argument gives unnecessary purchase to unscrupulous haters who might use that weaker argument as a low-grade "straw man" to dismiss the whole position.
Excellent point. Thank you for writing such a cogent argument.
It is clear that the rich and powerful are making short-term decisions that are causing widespread and long-term harm. Decisions such as, AI-pilled investors demanding unsustainable productivity from companies, and CEOs in turn forcing employees to use agentic coding and laying the rest off.
Such decision-making is not new, of course, but LLM technology is helping grifters and con artists to turbo-charge it. Founders can raise money using cool demos that achieve the impossible. Investment analysts can back up their AI story with huge paper profits.
The result, as you summarized, is life and career displacement and missed opportunities for young people. It's just too abstract and diffuse of an impact to influence people with, and it also seems to be railing against the holy cow of Capitalism.
"Data centers bad!", or "AI is disrupting humanity and the end is nigh!" are much more punchy slogans.
Programmers pervasively underestimate the sheer amount of manual brunt work that business users routinely do by hand and with their own eyes, mouse, and keyboard to make spreadsheet-based processes "work out."
I work in Finance. It's routine for the folks I work adjacent to to spend literal hours performing a task using a spreadsheet. The same logic could be executed by a query engine or scripting language in seconds. But... I know from trying... it would take months if not years to accurately identify and account for all the weird exceptions and edge cases that the human operator is accounting for intuitively because of their undocumented tacit knowledge of the domain.
So while the spreadsheet itself as an artifact is a mess, and the process as a whole is gallingly inefficient... the outputs they produce are often preternaturally correct.
I mostly agree. There are certain classes of tasks where the spreadsheet is more efficient. It's "more efficient" because it 1) requires less set up time (like you said, translating a computation in a more traditional system could take months), but also because 2) the spreadsheet is a highly interactive and accessible environment.
We have a programmer on our team who keeps trying to convert our excel calculations to more structured systems (database store). The issue is that the spreadsheet-work almost always requires navigating and making decisions based on unstructured data. The format isn't formally specified, so in the end our programmer makes more mistakes because he doesn't engage with the data - the processing gets treated as a routine algorithm, and it's not. The interactivity (and the fact that we're not spending energy on formal specification) is what allows us to more effectively engage with the data.
Another important point is that spreadsheets more easily allows you to consider data points individually without hand cuffing yourself to one kind of processing algorithm or software design. I can easily edit one cell to make a correction in a data set (which is a double-edged sword). The programmer has to create an additional handler to handle the exception. Depending on the design/quality of the code, this could just be adding undesirable complexity in the software. So we can either spend 30 seconds and edit a cell, or spend half an hour adding to code (which by the way may never get used again).
When I and my (still fairly young) family needed to move cross-country, my wife and I accepted a credit card offer with 0% APR for the first year and put all our moving expenses on it. Then once we settled, we paid it down a bit at a time each month, and then right before it would have started charging interest we paid the rest as a lump sum.
Really helped us float the moving company and also some DIY renovations on our house that we didn't have all the cash on hand to pay for outright. And we didn't pay a thin dime for the privilege.
In that podcast DHH was talking about how in Omarchy you can change the system settings, monitor setup, themes, window chrome, installed programs, window manager config, keyboard shortcuts, etc. using agents. This is facilitated by Omarchy because all that stuff is dictated by text files that agents can directly interface with.
Whatever the merits of MacOS, and there are many, it is not the paradigm there that all your system settings and config are stored in text files. That's what he was talking about.
Do you have a non-video, non-X summary of what he’s talking about?
I’m not watching a 31 minute video on X to figure out what this is about. Theo is another influencer who thrives on being contrarian, embellishing negative claims about things he’s not pumping right now, and trying to get everyone to ignore negative things about what he is pumping. I would not trust his judgment on topics like this without some third party benchmarks and deeper claims because the first priority of influencers in this category is to be contrarian and ride divisive topics. Being factually correct is secondary.
Ah, the irony of posting "Your Mac is slowing you down" on the bloated JS hellsite that is X. That page took 10s to appear on my machine (before even thinking of rendering the video) whereas the HN front page is <1s.
"it is not the paradigm there that all your system settings and config are stored in text files."
This is inherently untrue. A .plist in XML is nothing but text. It might be formatted text, but it is still text nonetheless, like almost every other configuration file in existence.
This is a theme in Blindsight by Peter Watts as well.
In that setting, field experts working at the bleeding edge are so advanced that non-experts literally can't understand what they're saying at all. So there's a whole class of specialists, "synthesists", that specialize in gaining approximate understanding of the experts' work for the purpose of communicating it to outsiders—perhaps wrongly, according to the expert at least, but hopefully more productively vs the unmediated version.
What's amusing to me in this context is, summarizing emails and such has for a while been a supposed use case for AI—the LLM serving as the "synthesist" to explain long texts accessibly. But with this math question, a human "synthesist" would be needed to approximately understand the math discovered and programmatically verified by the LLM. So the roles reverse.
If it isn't still common practice 40 years from now (8/18/2066), I'll give the first person to challenge me and cite this comment $1 USD (or equivalent value in the One-World Order-issued omni-currency that we will probably be using by then).
This sounds roughly similar to, "If eating steak is nutritious and tastes good, why would you not eat an 8oz sirloin for breakfast, lunch, and dinner for an entire month?"
"Too much of a good thing" does exist. Someone might feel that it takes time for their organization to "metabolize" the rapid additions made by coding agents. That seems quite reasonable to me. Your comment seems to argue a false dichotomy.
Some phones bypass the battery when plugged in, but not all do. I know it's advertised as a feature of some gaming phones because graphics-intensive gaming is one use case where you'd want to use the phone heavily while plugged in.
Steamdeck lets you set your preferred % and the dock does pass through charging/power, which I love because that combo guarantees the longest possible life for the battery. I turn it back to 100% for flights, otherwise leave it as 80% at all times. Mine is almost 4 years old now but I don’t feel even a hint of battery degredation yet.
Even then, some phones that advertise bypass charging are actually using a software-based implementation, and they can still end up with swollen batteries after being plugged in for extended periods.
One way of defining Wisdom is the ability to look at a set of metrics and discern when an adjustment or adaptation to them is needed in order to better approach the true objective we are optimizing toward—since the true objective is almost never perfectly described by any of the metrics that we can find to approximate it.
Eventually, almost every attribute we can practically and objectively measure is at best a proxy measure for the thing we actually want to know, which is, in this case, "Most likely to win races." The thing we actually want to measure is often hard, impossible, or impractical to objectively measure, which is why we resort to proxies in the first place. By "proxy measure" we mean, not a perfect 1:1 to the thing we want to measure, but containing some amount of signal about it.
So yes, sometimes there's a peak in the utility of speed. But perhaps that trade-off can be modeled by another attribute called "Proneness to fly off the track" which increases as speed does, and is weighted such that the peak and then taper-off of benefit in speed is captured between the combination of the two of them.
But then, any vehicle that has the same speed but lower "Proneness to fly off the track" would be objectively preferable to any equally-fast vehicle with greater "Proneness to fly off the track." And so it continues.
But I think it's important for the quality of public discourse that we forward only the best, strongest arguments for any given position (whether I personally agree with that position or not). A weak argument gives unnecessary purchase to unscrupulous haters who might use that weaker argument as a low-grade "straw man" to dismiss the whole position.
So with that preamble, some constructive criticism (very well-intentioned):
> I am not exaggerating: We need to resist “AI” so that thousands, perhaps millions of people do not die.
On this one you are, in fact, exaggerating (certainly with "millions," likely with "thousands" as well).
The overwhelming majority of people who lose their job for whatever reason (and there are many reasons, not just this one issue) fall back on a support network of their family and friends while they "get back on their feet" by pivoting to another job or form of employment (sometimes branching into entrepreneurship, for example).
The alternative to employment at whatever FAANG or FAANG-adjacent company someone was laid off from is not "death." It is typically "some, sometimes quite significant, discomfort." Granted "AI Layoffs" are starting to ripple out beyond FAANG, but the story is similar in other industries.
If "layoffs kill people" is the moral hill we choose to plant our flag on, then mustn't we also campaign with equal vociferousness against all the other causes of layoffs in general? Since to do otherwise is to sit by while those laid off allegedly die in droves of thousands or perhaps millions?
It also occurs to me that this framing ("AI causes layoffs") also gives incompetent and unaccountable executives a pass by pinning the actual harm done by irresponsible layoffs (which to be clear, is not none!) on "the AI" rather than on the leaders who are in the positions to actually decide what happens in their organizations.
I think the argument is much stronger if we talk about the aggregate cost to humanity being paid in terms of displacement, life disruption, and curtailed opportunity (especially for juniors). We should not make the "toxic escalationism" move of equivocating things we don't like with negligent homicide, even if it makes for a punchy one-liner.
reply