Maybe you’re comparing it to some idealized view of what human therapy is like? There’s no benchmark for it, but humans struggle in real mental health care. They make terrible mistakes all the time. And human therapy doesn’t scale to the level needed. Millions of people simply go without help. And therapy is generally one hour a week. You’re supposed to sort out your entire life in that window? Impossible. It sets people up for failure.
So, if we had some perfect system for getting every person that needs help the exact therapist they need, meeting as often as they need, then maybe AI therapy would be a bad idea, but that’s not what we have, and we never will.
Personally, I think the best way to scale mental healthcare is through group therapy and communities. Having a community of people all coming together over common issues has always been far more helpful than one on one therapy for me. But getting some assistance from an AI therapist on off hours can also be useful.
I've actually gone in the other direction. A year ago, I had that feeling, but since then I've gotten more certain that LLMs are never going to be able to handle complexity. And complexity is still the real problem of developing software.
We keep getting more cool features in the tools, but I don't see any indication that the models are getting any better at understanding or managing complexity. They still make dumb mistakes. They still write terrible code if you don't give them lots of guardrails. They still "fix" things by removing functionality or adding a ts-ignore comment. If they were making progress, I might be convinced that eventually they'll get there, but they're not.
Yeah but on the other hand there are plenty of human programmers that are bad at understanding complexity, make dumb mistakes, and write terrible code. Is there something fundamentally different about their brains to mine? I don't think so. They just aren't as good - not enough experience, or not enough neurons in the right places or whatever it is that makes some humans better at things than others.
So maybe there isn't any fundamental change needed to LLMs to take it from junior to senior dev.
> They still "fix" things by removing functionality or adding a ts-ignore comment.
I've worked with many many people who "fix" things like that. Hell just this week, one of my colleagues "fixed" a failing test by adding delays.
I still think current AI is pretty crap at programming anything non-trivial, but I don't think it necessarily requires fundamental changes to improve.
this whole analogy is so tired. "LLMs are stupid, but some humans are stupid too, therefore LLMs can be smart as well". let's put aside the obvious bad logic and think for one second about WHY some people are better than others at certain tasks. it is always because they have lots of practice and learned from their experiences. something LLM categorically cannot do
> LLMs are stupid, but some humans are stupid too, therefore LLMs can be smart as well
Not what I said. The correct logic is "LLMs are stupid, but that doesn't prove that they MUST ALWAYS be stupid, in the same way that the existence of stupid people doesn't prove that ALL people are stupid".
> let's put aside the obvious bad logic
Please.
> WHY some people are better than others at certain tasks. it is always because they have lots of practice and learned from their experiences.
What? No it isn't. It's partly because they have lots of practice and learned from experience. But it's also partly natural talent.
> something LLM categorically cannot do
There's literally a step called "training". What do you think that is?
The difference is that LLMs have a distinct off-line training step and can't learn after that. Kind of like the Memento guy. Does that completely rule out smart LLMs? Too early to tell I think.
That's a misunderstanding. I argue elsewhere in this thread that these categories aren't real, yet I identify as neurodivergent.
One problem is that the language we've been given is one of defects (ADHD, ASD) and often people are stuck using that label even if they don't see their trait as a defect. So we're stuck saying "ADHD" because psychiatry decided that it's a defect. In reality, nearly every "disorder" has some other context-dependent benefit.
And while there's no "neurodivergence" in nature, there is a divergence from what is considered "normal" by society. "Normal" is an artificial construct, but we all have to live with it, and there are many aspects of my nervous system that clearly function in a way that's outside of two standard deviations from the mean, and this causes me a lot of problems in a society that can't tolerate that. These are very very real, and 95% of the time, I never say a word, and no one else has any idea that I'm suffering. I'm not telling anyone else that they're the asshole. Most neurodivergent people are living like this, despite what you're implying, that we're all just walking around being assholes to everyone we meet. That, is bullshit.
> Autism exists, to the extent that any psychiatric disorder exists.
Which is to say, not really. I say this as someone who has been diagnosed as autistic, and identifies as autistic. All of these diagnoses are presented as clear, well defined constructs that exist in the world, but in reality they’re fictions that that committees have drawn around a vast gradient of human traits.
No individual human truly fits any single diagnosis. For example, I have two family members that depending on how you frame their behaviors could be described as either autistic or narcissistic, yet these are supposedly completely different disorders. Prior to being diagnosed as autistic, I’d been diagnosed with some of the ones suggested in the article as well. Was I misdiagnosed? I don’t think so. None of those constructs are real either. So, they’d not even wrong. For a time, some were useful. Some were harmful. But seeing myself as autistic has been a lot more useful.
What matters to me about identifying as autistic is that it allowed me to find other people who experience the world similarly to me. Until I found other autistic people, I felt like I was a single alien stranded on Earth, alone. Finding other autistic people was like finding out that there were millions of other aliens like me hiding in plain sight.
I hope that someday we can move beyond the 1950s-style nosology of the DSM and have a more rigorous science of mental health, but right now, it’s what we’re stuck with.
I am sorry, but if you’re saying there’s no biological, physiological or neurophysiological evidence of these conditions then you’re just plain wrong. I cannot emphasize that enough.
That’s not what GP is saying. He’s saying that a term like “autism” is a lasso trying to capture a gigantic number of individual traits and symptoms. This is true of any other “psychiatric disorder” as well. There is no “autism”, there is no “ADHD”, there is no “OCD”, any more than there are tables or chairs.
Something being a table is a label we slap on it to abstract certain attributes, that allows us to reason about it without having to think about all of the non-table-attributes it has. What do tables do? What can we do with them? We can put things on, eat off them. We can’t feed them to our pets. We can’t use them as a trampoline. The object being “a table” is just a categorization we make to allow us to think about the object; it isn’t something that the object is.
Similarly, people aren’t “autistic”. They’re just people, who have certain traits, which psychiatrists have decided should be lumped into a category called “autism”, because they’ve noticed a cluster of other people who have similar traits. So, from this standpoint, someone “being autistic” does not tell us anything. We can already see that person’s traits or characteristics. That categorization might be helpful to some people, and it might be harmful to other people; and they should use or avoid using it accordingly. But they can choose to do that, because “autism” isn’t a “thing” - it’s a mental construct.
It’s the same thing as any condition which deviates from the set of characteristics considered “normal” for a given population.
Eczema is a skin condition which happens to some people, it’s not something that happens in most people. But we can see evidence of varying degrees of severity of skin damage due to eczema. This condition can happen for any number of reasons, immunological, endocrinological, or some combination of factors. There are different types of eczema, but for ease for conversation with anyone other than a doctor, you just say you have eczema.
Same for mental conditions, they have their underlying causes, and some representative characteristics we found on average and grouped them as classes for ease of diagnosis and treatment.
I understand the folly of mischaracterizing, so it doesn’t make sense for researchers or medical professionals to not care about the categorical distinctions.
However, as far as the normal public is concerned, someone’s problem is their problem, and they don’t owe you a detailed explanation of their condition, or a doctors note because you’ve been socially offended (I understand maybe that’s not the point in either of your posts, but I thought I should say it now that it occurred to me in the flow of this post).
That tend to hold "on average" for a population but often don't hold for the individual within a population. This is the ecological fallacy [0], just one of the fallacies underlying psychiatry.
My argument isn't that psychiatric symptoms don't exist or aren't real and there is no real underlying phenomenon. My argument is simply that we've drawn the lines between the units of study too high up and we should be more granular. This level of nosology was chosen in 1952. Do you really think they got it 100% right almost 75 years ago? And what is the mechanism for defining and maintaining these categories? A bunch of committees get together every few years and decide on them, then they tell us all what's "true". Bullshit. What are the odds that a committee will define itself out of existence? Pretty slim. [1]
I have traits that could be considered as autism, ADHD, obsessive compulsive personality disorder, PTSD, bipolar II, social anxiety disorder, and probably a dozen more disorders. But by quantizing the disorder at the current level, by necessity, the other traits are cropped out of view. Relevant information is lost and irrelevant information is blurred together. And the level of overlap between disorders is absurd. They cannot possibly be "real" because the lines between them aren't even distinct.
The useful unit to study is the individual trait, not the cluster of traits that is different in each individual. The traits are more granular and map more closely map to underlying biology anyway. The current model is akin to what the geocentric model was in astronomy. It's outdated, wrong, and holding us back from a more accurate, detailed view.
> My argument is simply that we've drawn the lines between the units of study too high up and we should be more granular.
I agree with this, and your overall post. I’ll just add that if the purpose is treatment, it helps to find root causes, and maybe there’s a common thread in the underlying root causes, likely related to gene expression.
Perfect is the enemy of good. HTML is good enough. Let’s get this done.
And as another commenter has pointed out, HTML does exactly what you ask for. If it’s done correctly, it doesn’t contain font sizes or layout. Users can style HTML differently with custom CSS.
mixing rendering definitions with content (PDF) is something from the printer era, that is unsuitable for the digital era.
HTML was a digital format, but it wanted to be a generic format for all document types, not just papers, so it contains a lot of extras that a paper format doesn't need.
for research papers, since they share the same structure, we can further separate content from rendering.
for example, if you want to later connect a paper with an AI, do you want to send <div class="abstract"> ... ?
or do some nasty heuristic to extract the abstract? like document. getElementsByClassName("abstract")[0] ?
All of the interesting LLMs can handle a full paper these days without any trouble at all. I don't think it's worth spending much time optimizing for that use-case any more - that was much more important two years ago when most models topped out at 4,000 or 8,000 tokens.
I went through the same pain with React Router. I would never consider using Remix or anything else from this team. It doesn't matter how great it might be or how much I might agree with the architectural principles. Who knows what shiny new idea they'll be chasing in Remix 4? I'd have to be crazy to trust them.
> I had to read this over a few times to believe that I was seeing it. If it didn’t include the word “literally” I’d assume some poetic license on the part of whoever, or whatever, wrote this.
Unfortunately, the meaning of the word “literally” has morphed into almost the opposite of “literally”. Most people just use it as an intensifier devoid of any true meaning. Makes sense that an LLM that doesn’t have any sense of truth would just stuff that in there.
HN is heavily biased towards silencing any content that makes any subset of its readership uncomfortable. This policy exists under the mistaken (I hope) assumption that silence is not fundamentally biased, which is usually not the case.
Posts about climate change, Israel/Palestine, uncomfortable concerns about the state of the industry etc will very often disappear quickly once they hit the home page.
Of course silence about climate change is beneficial for those contributing the most to it, silence about Israel/Palestine is very beneficial to Israel, and clearly silence about lack of concern for child safety by major tech companies benefits those companies.
The real-world conditions are beneficial to israel. I don't think yapping about it will resolve the conflict any more than it did for tibet.
What more can be said about the matter? Some pro-palestine organization will accuse Israel of genocide, and Israelis will continue bulldozing inch by inch.
As for "uncomfortable concerns about the state of the industry", those are on the front page and top of /active regularly
> HN is heavily biased towards silencing any content that
... violates the guidelines, in particular:
> Be kind. Don't be snarky.
> Comments should get more thoughtful and substantive, not less, as a topic gets more divisive.
> Please don't fulminate. Please don't sneer, including at the rest of the community.
> Please don't use Hacker News for political or ideological battle. It tramples curiosity.
> Please don't post insinuations about astroturfing, shilling, brigading, foreign agents, and the like. It degrades discussion and is usually mistaken. If you're worried about abuse, email hn@ycombinator.com and we'll look at the data.
We have a culture where we’ve been told for decades that market forces and the profit motive are sufficient for running a society. That the market will find a way to give everyone what they need efficiently without problems.
We’ve dispensed with ethics as a basis for human interaction, and the results are exactly what one would expect: a dystopia.
And the people making the most money off this system insist that it’s all for the best and that we should double down on this strategy. Any mention of putting limits on greed and exploitation is met with responses like, “what are you, a socialist?” as if the only two choices for structuring a society are either a rapacious hyper-exploitative capitalism and an oppressive Soviet state, and there’s no other option.
Capitalism needs constraints. Capitalism in the service of society can be a great thing. Capitalism without constraints is a cancer that will destroy everything in the pursuit of profit.
It's hyper individualism that is fueling this type of capitalism, the notion that you owe nothing to anyone and other people in society are not your concern, society itself is not your concern, it’s only about what you want etc.
It's like a pond, if you follow hyper individualism, you will extract as many fish from the pond as you can, without caring what happens next or others using the pond, and the fish will not be able to reproduce for future generations, others will not be able to get as many fish as they need etc.
It needs to be balanced. As you say, there is something in between that can work, we do not need to choose only between extremes.
If humans were much more rational this would work better.
The human brain is loaded with exploits, and capitalism being an excellent optimizer quickly finds and uses these exploits. Because they work, and more importantly they are way way easier than creating actual value.
A casino is more profitable than a hospital. Quack medicine sold with sensationalism is more profitable than real medicine. Porn is more profitable than good film or literature. Rage inducing click bait is more profitable than actual news or thoughtful editorial. It’s kind of just thermodynamics. These things require less energy input, and they don’t have to “work” because they exploit security vulnerabilities in the dopamine system instead.
> We have a culture where we’ve been told for decades that market forces and the profit motive are sufficient for running a society. That the market will find a way to give everyone what they need efficiently without problems.
I don't really think that culture has existed lately, it kind of died out with the 2008 financial crisis. Now it's about naked use of power, whether political or economic.
The problem with constraints on individual freedom (which is essentially what happens when you constrain capitalism) is that no one agrees on what they should be, and therefore a segment of society will not be happy with them, and claim them as tyrannical oppression. Sometimes this is hysterical nonsense, sometimes it has a point.
Ultimately the antidote to unfettered capitalism is sensible policy crafted through political compromise. But largely Western politics itself has skewed towards extremes lately, few have the patience or understanding for this process, they want a quick fix.
The crony capitalism was brought to you literally by companies and millionaires scared of regulation and heavily pushing for deregulation. They supported Trump and republicans by large margin, especially financially.
If you think only one of the two parties of capital are to blame for this timeline, I have a bridge to sell you.
Democrats manufactured consent for Trump's rise in a 1000 ways over decades. Neoliberalism delivered us Trump, who is merely a symptom of this broken system.
Dems/Repubs play good cop/bad cop; always in service to capital interests. We have a uniparty, but people are fooled by reasonable election turnout and close elections. Democracy is an illusion in the US.
Yeah, you know what, the culpability of republicans is so much higher at this point, that yeah, they are the primary responsible for current state.
This both sides game is just how they were enabled again and again by people who like to frame themselves as reasonable above the politics center, but are in fact just their supporters.
I don't think any actual socialist would ever argue for an oppressive Soviet state either. They'd want stuff like public firefighters, health care, sewage, roads, etc.
We need to stop painting with the wide brush of "capitalism" like this.
This is not capitalism vs socialism. You are describing neoliberalism.
A good place to start is the Wendy Brown book Undoing the Demos. The subtitle is "Neoliberalism's Stealth Revolution"
There is no way to change and fight back against something when we are so confused what we are even fighting against.
No one is really against market prices. They are against these insane, distorted, neoliberal ideas that apply the efficient market hypothesis, "the market is always right" to non-markets. Or using the efficient market hypothesis as a pseudoscientific moral justification for bad behavior.
Most importantly is that there are a ton of people that actually believe this bullshit.
If a process causes number to go up then the process must be morally good because "the market is always right". Obviously, if the process was morally bad number go down because "the market is always right". This is the 21st century American religion.
Railing against "capitalism" just causes those hypnotized by neoliberalism to completely tune out before you even get to the ism in "capitalism".
Why? Most large corporations I’ve dealt with are highly bureaucratic and resistant to change. Good ideas get lost in silos or bogged down in bureaucracy. Whether it works or not seems entirely dependent on whether the company has a moat around their revenue stream, which allows them to be inefficient everywhere else.
For an employee owned co-op, a more anarchistic organization structure that allows for more employee control of everyday decisions could actually allow the company to adapt and change more easily. The ones making decisions have skin in the game, both as workers and owners.
If the employees have the power then decisions that are good for company but bad for the employees won't be made.
Let's say that the company can't compete so the CEO proposes to automate production and lay off 50%+ of the employees, do you think employees will vote in favour?
In general coops are not good at tough decisions and innovation.
Duralex already went bankrupt several times and they are heading for it again. What's in the article is nice but it's charity not business so unfortunately I am not optimistic.
Were they a coop when they went bankrupt? According to the article, they only became a coop recently, so having a CEO capable of firing everyone didn’t work out for them.
They took investors, who agreed to the rate of return on their investment. That doesn’t sound like charity.
Now, your example of a CEO that wants to fire everyone assumes that that’s the right decision. How well has that kind of thinking worked for other firms like Boeing? That type of authority structure introduces its own set of distortions, which usually skew towards shareholders, and often not towards long term sustainability.
As a worker, I would be against that decision for selfish reasons as well as for rational reasons. It sounds like a bad idea. If they want to sell commodity glassware, then that’s a race to the bottom. But they’re selling quality, which requires humans with skill.
> In general coops are not good at tough decisions and innovation.
This needs to be backed up. Mondragon in Spain has thrived for decades. In America, mutual aid societies used to provide health care, unemployment insurance and other benefits before being squeezed out by other groups who were better at things like regulatory capture.
There is a long history of cooperative ownership that goes beyond the stereotypical hippie grocery store. I think it’s too quick to dismiss Duralex.
They were not. In fact they went through series of mass-redundancy episodes that were supposed to save them from bankruptcy, soon followed by yet another bankruptcy.
The COOP might fail. Indeed the call for contributions discussed in the article was motivated by that risk. But it won't fail because it was a COOP, because every CEO who tried also failed to save it. The COOP structure is this company's last chance, literally.
Their Wikipedia page says that headcount is flat to increasing over the last 10 years.
The coop structure is a result of their bankruptcy in 2024. As I understand, this was the proposal that didn't involve any layoffs and it was chosen by the bankruptcy court. They also got a large de facto subsidy. The fact that they have run out of money again so quickly (the root of the article) is quite worrying to say the least.
None of that changes anything to the points of my previous comment.
Except that these employees had been put on temporary unemployment (chômage partiel) [0] and their difficulties have been going on for much longer than 10 years.
Ultimately all I am saying is that other structures have not fared better than the COOP. Claiming that the potential current failure is because the COOP prevents hard decisions to be made while ignoring the fact that the previous owner lasted 3 years before failing (in spite of the temporary unemployment decisions) is not logical.
I have never claimed nor suggested that their situation is because they are a coop...
Most of the replies I got here have not even read my comments, apparently, and completely beside the point or just rush to condemn me for blaspheming. It's like the poor guy who dares disagreeing at a student socialist meeting.
Here is a quote from your original message which I have read and suspect others have too:
> In general coops are not good at tough decisions and innovation.
> Duralex already went bankrupt several times and they are heading for it again. What's in the article is nice but it's charity not business so unfortunately I am not optimistic.
These two paragraphs following each other do make it seem like you are making a connection between the two, coops being unpropitious for hard decisions and this particular coop heading for bankruptcy.
That original comment was primarily made up of three paragraphs criticising coops, and readers naturally assume that the final concluding paragraph (this place is likely going bankrupt) is linked to the first three.
> just rush to condemn me for blaspheming.
Nobody has accused you of blasphemy. They just disagree with you. You are not being victimised, and nobody has pretended you were not allowed to think as you think. There is just a discussion taking place between disagreeing people.
> Duralex already went bankrupt several times and they are heading for it again. What's in the article is nice but it's charity not business so unfortunately I am not optimistic.
Everyone can rally round in their time of need but that doesn't change the fact that Duralex was struggling to begin with, and once this goodwill windfall dries up they'll be back here.
> If the employees have the power then decisions that are good for company but bad for the employees won't be made.
"The company" is a fiction. Real parties are, e.g., employees, capital owners, suppliers, and customers (and, to the extent that any of those are corporations or similar convenient fictions, the same kinds of groups with respect to those entities.)
With a pure labor coop (which an SCOP isn't quite, put similar enough that it works as an approximation for discussing general traits), "capital owners" and "employees" are the same group, rather than different groups whose interests are frequently adversarial.
> Let's say that the company can't compete so the CEO proposes to automate production and lay off 50%+ of the employees, do you think employees will vote in favour?
Quite possibly, though because of this exact issue (and generally the need to buy out ownership shares of terminated employees), labor coops are less likely to go on hiring binges that force them to rapidly and massively downsize to survive when the conditions that drove the binge change.
Maybe you’re comparing it to some idealized view of what human therapy is like? There’s no benchmark for it, but humans struggle in real mental health care. They make terrible mistakes all the time. And human therapy doesn’t scale to the level needed. Millions of people simply go without help. And therapy is generally one hour a week. You’re supposed to sort out your entire life in that window? Impossible. It sets people up for failure.
So, if we had some perfect system for getting every person that needs help the exact therapist they need, meeting as often as they need, then maybe AI therapy would be a bad idea, but that’s not what we have, and we never will.
Personally, I think the best way to scale mental healthcare is through group therapy and communities. Having a community of people all coming together over common issues has always been far more helpful than one on one therapy for me. But getting some assistance from an AI therapist on off hours can also be useful.
reply