Tons of corporations rely on the fact that human interaction within software production will be as minimal as possible, and no one actually cares who trains the seniors of the future.
A parent doesn't necessarily knows best or acts with your best interests in mind, and that's really observable by all the toxic parents regardless of whether there's malicious intentions behind their actions or just sheer incompetence.
I'm too European for the rest of your argument and I just happen to be Greek so I fundamentally disagree with the premise that I'm voting a father or mother figure which I must blindly follow out of obligation or blind trust.
>A parent doesn't necessarily knows best or acts with your best interests in mind..
A parent who does not have the best interest of their offspring is a fool..Now the "knowing best" part. I have a news for you. No one does, unless they have access to a crystal ball that shows them the future..
> toxic parent
As soon as I see the word "toxic" in a comment, I lose all respect for the comment.
> I'm voting a father or mother figure which I must blindly follow out of obligation or blind trust...
No one is saying you should trust them. What my comment is saying is that you cannot declare how much a person care by looking at external behavior. Even if one put a gun to someone's head and blow their brains out, if you had more context, it might turn out to be in their best interests. (Reference: ending of the movie Mist)...
That's just awesome and makes me regret not switching to Linux earlier. I wish the creator of this made the start menu suddenly change after running Windows update, because Microsoft knows best on which folders you need to be visible by default. It can be customized, but you can't go back to how the start menu was.
Someone who knows that 1 + 1 = 2 will not decide that it's suddenly 3 unless we start accounting for health problems. Making mistakes is not the same as non-deterministic.
Lol, humans make such absurd mistakes (and worse) all the time through simple typos, which is effectively random. The key for 2 is right next to the key for 3, after all.
I think that actually reinforces the distinction being made. An LLM’s nondeterminism is in the generation process: given the same prompt and model state, sampling can produce different outputs. That doesn’t mean the underlying fact itself becomes nondeterministic.
A human who knows 1+1=2 can still say “3” because they misread the question, misspoke, were distracted, or made some other cognitive error. Likewise, an LLM can output “3” because the generation process selected an incorrect continuation. Those are both errors in producing an answer, not evidence that 1+1 somehow has multiple answers.
So yes, human mistakes and LLM sampling are mechanistically different. If your argument is that LLMs and humans can both make mistakes, then major question here is why are we building out huge amounts of infrastructure at unsustainable spending levels to enable LLMs to make the same mistakes as humans.
> If your argument is that LLMs and humans can both make mistakes
It's not, I'm just pointing out that LLMs won't make that mistake.
You could ask an LLM what 1+1 is, and the number of times it says "3" is so small that it makes no sense to worry about it. It will phrase the response differently each time; that's the nondeterminism. But it won't say "3".
> then major question here is why are we building out huge amounts of infrastructure at unsustainable spending levels to enable LLMs to make the same mistakes as humans.
Yes, if we ignore everything else, that seems like a reasonable question. But let's not ignore everything else, like the fact that LLMs are much more productive than humans and likely already make fewer mistakes than the average programmer.
>You could ask an LLM what 1+1 is, and the number of times it says "3" is so small that it makes no sense to worry about it...
I think the disturbing fact is that you can take a frontier model with all the intelligence of humanity, and make it say 1 + 1 = 3, by specifically training for it...
A human with that much knowledge will refuse that attempt. There in lies the difference..
Yes, I fail to see anything meaningful. If you move the goalposts and say “I have invented a human that cannot be convinced in any way to give a wrong answer” then what’s the point of that in this discussion, really?
And you seem confused about how an LLM works and what it is - “the intelligence of all humanity” - not how it works. You’re debating using 2 imaginary things you created.
Hit them with a stick until they answer as you told them to.
I think the post up-thread, https://news.ycombinator.com/item?id=49881653, was trying to make this point by linking to an episode of Star Trek TNG, with Picard being tortured until he said the "correct" (incorrect) number of lights.
(I recommend against using fiction as evidence; in this case the general point happens to be valid, and is why torture is forbidden: we humans really do break, but breaking doesn't mean we tell the truth, it means we tell people what we think they want to hear).
> Obviously, for this purpose, human should not have any feelings (because LLMs don't have), so can't feel pain. Or else the comparison can't work.
Other than this forcing you to ignore the overwhelming majority of humans who have functioning pain nerves:
LLMs have something functionally equivalent to pain, in this regard at least.
During training, model weights are updated depending on if the feedback was positive or negative.
It has a functional effect similar to that which pleasure and pain have with us. Not identical, so far as I know there's not been any reports of any machine learning model that is into BDSM, but for the most part functionally similar.
There is also research which has found circuits in multiple LLM models, which serve similar roles at inference time and are distinct from other emotional representations: https://arxiv.org/abs/2609.16247
> Someone who knows that 1 + 1 = 2 will not decide that it's suddenly 3 unless we start accounting for health problems.
And?
The p(that kind of error) is pretty small now. At what point does a probability coming out of an LLM look like "knowing", such that spitting out the wrong answer despite that probability looks like a health problem, a typo, or even just boredom? (Thinking of the Lizardman constant here: https://en.wiktionary.org/wiki/Lizardman%27s_Constant)
It's a continuum for both them and us, even if the mechanism is wildly different.
> Making mistakes is not the same as non-deterministic.
i.e. when the dismissal is "non-deterministic" when it should be "Making mistakes", is itself a mistake.
> Someone who knows that 1 + 1 = 2 will not decide that it's suddenly 3 unless we start accounting for health problems.
But this is plainly false. This kind of unforced error occurs all the time.
For example, once when I was in high school I traced an error in my math homework to an intermediate calculation of "2 + 2" as being "3". There was no reason.
What we can say about humans is that, if they know that 1 + 1 = 2, (a) they are unlikely to change their mind about this in any kind of lasting or permanent way, and (b) the rate at which they will mistakenly produce other values for 1 + 1 is very low. But it will happen occasionally, and when it does happen, "they just suddenly decided on the wrong value" is an extremely accurate description of what that looks like.
> It is absolutely wild to me how a corporation can be so toxic and abusive to its users and still be in business.
It's easy: The choice was made from someone else, and you just accepted it as normal. Hardware vendors shipping PC's with Windows 11, corporations who went all in to Microsoft's tooling shove Microsoft Teams down your throat because "the company uses it", and the list goes on.
I recently found out that several researchers, departments... buy a personal subscription for zoom (out of the own project money) even as our university as a MS Teams subscription. Despite if it is effectively "free" for them, they do not use it. So much on public money spend wisely and the quality of the MS Teams software...
Can confirm, my boss has his own paid Zoom while I’m using Teams to save research money. Another assistant professor I know pay for Slack as well.
I think Teams improved and Zoom regressed in the last 3-4 years to the point the garbage level is not that different nowadays. Most of the annoyance is related to finding your way in a different interface.
Avoiding Office 365 is so annoying due to everyone in academia using it that I’m using macOS instead of Linux at work just to interface with others better. I got used to it but I need Linux as soon as I’m home, I feel observed and vulnerable otherwise.
LaTeX absolutely sucks for proofing. Typos end up being buried in tags, and you end up with a document that looks pretty but possibly not fully edited.
If you are collaborating with advisors, they may wish to add comments to your copy as you work through your thesis. There is no clean way to do this in a TeX source file, so the only way to do this would be to export your work-in-progress into a DOCX, and then work your changes back into your source.
Versioning-by-hand sucks balls. When you’re approaching various deadlines with other obligations at hand, all open-source kumbaya ideas go out the window and you just make do.
For your manuscript_final_FINAL_v2.docx file, manual layout tuning ends up being not too bad and likely takes less time than troubleshooting LaTeX which might not have as much documentation.
Last time I wrote anything remotely academic was several decades ago, but at that time, the argument was that you had to use LaTeX because Word documents had no support for simultaneous editing, version control, and editors comments. It is truly bizarre to see the same arguments repeated after so long time but completely reversed, in favor of Microsoft Word. Surely the tools haven't become worse over time, given the popularity of github, Overleaf and such?
Don't most people use git for their latex (or even markdown) version control? All the tools that have been developed for code collaboration work really well for documents compared to messing around with word's undocumented format and binary blobs.
Disagree. Hunspell and aspell have tex modes that understand well what needs to be checked and what doesn’t.
Comments can be added with ‘%’. Just make sure to have a linebreak at every sentence and it’s cleaner than anything. Services like overleaf also exist for collaboration.
You can use any versioning system you want. From git to dropbox, its plaintext.
I do agree that layout can be very time consuming, but it helps once you learn to not care about it till the absolute end of working on the content.
It isn't realistic to teach your collaborators a new system just for your own comfort. They'll just avoid making edits/comments. Especially if you're just a PhD student/postdoc interacting with a full time scientist - who will have much more work to do than fiddling around with your esoteric editing system.
It's kind of the same issue that FOSS enthusiasts often miss regarding other things, eg. most gamers wouldn't waste their time fiddling around with WINE settings to get things to run in Linux, the solution wasn't to tell them to just install something extra and do things different, it was to improve the infrastructure such that very little fiddling is needed in most cases.
Sure, I did my phd back in the 2010s and latex+git/svn was what everybody used. Im not saying you cant use word, im just saying the criticism raised of latex above is not a valid reason to.
LaTeX used widely for papers, but it does not cover all. PowerPoint is huge, Word is used closer to Markdown’s level for simpler documents (and Pandoc is a huge helper). All administrative staff is using MS Office, that means Excel is used a lot for forms as well. Then there is OneDrive and Teams for collaboration.
My experience has been that there's somewhat of an age divide. I've noticed that fellow junior academics are far more likely to prefer LaTeX, while all senior collaborators I have ever had, preferred Word. So I end up using Word for everything to ensure that it is easy for them to edit/review the paper.
Word is good for collaborative editing, but not very convenient for version control (and version control is very important when using AI assistance). So lately my strategy is to draft in Markdown, then copy over to Word once the draft is at a point that I might be okay sharing it with collaborators.
Not so much autocorrect, but Excel is very eager to interpret things as a date, which may be useful when you're dealing with dates a lot, but it disqualifies it for anything related to serious data.
People also used it to collect COVID data during the pandemic, and a lot of data was lost because someone accidentally opened it in an older version of Excel.
It's one of my pet peeves. Excel should not ever be used to maintain data. It's a fancy calculator.
Even if you write your thesis in latex, all interaction with the administration, grant applications etc. still happens in word documents. Thank God/John MacFarlane for pandoc.
Not sure why you'd not just use discord if cost is that much a factor, it has better calling that slack and ironically given what slack is used for way better code embedding etc.
I took online classes in both Zoom and Teams. I prefer Zoom for that because I can hide most of the UI (and the other people) to focus on the teacher and their shared screen
a lot of companies migrated to teams purely because it was already included in the subscription, IIRC they were forced to take it out by EU in 2023 for being anti-competitive.
No argument there, but annoyingly since they killed Skype (it’s own lovely pile of garbage and spyware), the only thing my parents will use to chat online is Teams.
I have tried to get them to use Discord or pretty much anything else, but they have been unreceptive.
I am on Linux and so I just use the browser version, which at least sandboxes stuff for me, I think.
My wife and father-in-law used to use Skype all the time, and when it shut down, the stars aligned as I suggested they try Signal and they both loved it.
Well of course I wouldn't expect corporations to know what ethics or human decency is even if it stared them straight into the eyes, but I would expect governments to establish laws and regulations such that the psychopaths wouldn't be able to abuse people to no end.
The quality of the government reflects the quality of the majority.
The lowering of the quality barrier of entry to power for the politicians enabled by the glorification of stupidity via the megaphone of social media and the resultant divided electorate has given rise to the assholes nakedly profiting from it.
No matter how toxic you are, you can amass support at the speed of light.
While I agree that the quality of the government is correlated with the quality of that government's people, you do also have other factors, namely the way the government is set up, which matter.
Take for example the USA. The entire electoral system is "winner takes all".
Duverger's law [0] tells us that this will lead to a two-party system, as opposed to proportional representation, which tends to create truly multi-party governments.
I believe many of the current problems in US politics are in fact side-effects of the winner takes all electoral system.
Each election, the only two choices are "candidate chosen by Republican Party" or "candidate chosen by Democratic Party". As such, elected officials are basically forced to first convince the members of each party to support them, which often leads to Trump-like strategies of "alienate half the country in order to please the other half", because that's a good strategy to get the party nomination, which is, in a two-party system, is basically a necessary condition to get elected.
On top that, news media and social networks have figured out that tapping into the human tribal instinct and framing politics as an 'Us vs Them' conflict sells more ads then being neutral, so the media ecosystem will tend to aggravate this two-party system.
It’s not directly causal, there is a feedback loop. The government has been failing to address the people’s needs for decades because the people failed to hold them to account. But the economy was decent enough for people to ignore this, except for occasional hiccups like 2008, which was papered over with QE and some crumbs for the masses (rebranded Romneycare under Obama).
Meanwhile educational standards have continued to decline, and this will inevitably affect both the electorate and their decisions. Education has been on a downhill trajectory ever since Roger Freeman convinced Reagan that free higher ed was creating a dangerous “educated proletariat.” This was the genesis of the privatization that created diploma mills, insane tuition costs, and the student loan crisis. Public schools gradually moved away from the humanities toward a narrow STEM focus, despite these fields not being a good fit for everyone. Coincidentally, a narrow STEM focus doesn’t give future voters a comprehensive and historically grounded understanding of government. (This has not been helped by left activists trying to overcorrect by rewriting history.)
Pressure has been building, and it found an outlet under Trump. Some steam has been released, Trump has 2 years left (maybe) and only die-hard supporters still believe in him. There will be a small window to either start fixing problems or set up a comprehensive police state to rule without public consent.
That's not true. Trump no longer has a majority (see the coming elections) yet continues to be of inferior quality at all times. MAGA no longer supports Trump as they did two years ago, he lost million supporters already, and continues to lose more. He is now a sinking ship.
> No matter how toxic you are, you can amass support at the speed of light.
> That's also not true either. Toxicity leads to backlash.
Wait till AI driven job losses really start to mount. Nothing divides people faster than an empty stomach. The republicans are leaving a poison pill - they will profit from AI and when democrats win and the job losses really start to hit hard, the democrats will be left holding the blame bucket. Obviously they are not exactly crying hoarse for AI safeguards/job protections either because national security/donors etc.
Currently it looks more and more like goverment itself consists of a lot of psychopaths who want to abuse people. Like abusing people is the main reason people go into politics. And then voters select those who want to abuse the outgroup, but don't know that politicians consider everyone as underclass, therefore even their own voters are ripe for abuse.
One of the most telling quotes of our political era is this one from 2019:
“I voted for him, and he's the one who's doing this. I thought he was going to do good things. He's not hurting the people he needs to be hurting.”
They voted for him to do “good things” where the expectation was he’d hurt the right people. This person stated it a little more plainly than most but I think this view explains the last 10 years or so.
It is possible to argue that basically all modern governments are inherently "abusive", in the sense that, to my knowledge, all modern governments will use the threat of violence to extort wealth from the population, in the form of taxes.
Now, in theory, these taxes are spent for the benefit of the taxpayers, in a philosophical tradition that can be traced back to at least the French revolution and the Declaration of the Rights of Man and the Citizen [1], and possibly earlier (the "no taxation without representation" [2] of the American Revolution, possibly other examples I'm unaware of). However, in practice, the ruling class has been known to help itself to that sweet, sweet tax revenue [3][4], and if we look at history, governments such as the Bourbon monarchy/Ancien Regime [5] whose claim to legitimacy rested not on "the will of the people" but "the divine right of kings" would do things like extort money from the populace to pay for foreign wars merely because their belief system held that the will of God was that everyone did whatever the King wanted, whether or not it appeared to be good for them.
We can compare modern governments with modern corporations - both will take your money and in return give you a service (you pay taxes and get police + an army to protect you, courts to resolve disputes, etc, you pay a Spotify/Netflix subscription and get access to streaming audiovisual content), but the modern corporation can't really force you to take that deal - Spotify can't put a gun to your head and force you to pay your subscription, while the government does force you to pay your taxes.
As such, if you believe that any transaction that is not freely consented to by both parties is abusive, then all modern governments are, by virtue of forcing people to pay taxes, inherently abusive.
Every government has taxed its population, and past governments often used real violence to collect them. The level of taxes we see currently is abnormal and abusive imo, but taxing in general isn’t.
Corporations need to deliver lots of value to their customers to get somewhere in the first place. If you want power it's much easier to become a politician and just blame corporations and spend the money they make and take the credit.
I don't want to challenge people's deeply held faiths, but I don't think we need the silly "maximise shareholder value" chant in every single conversation thread.
I am sorry, but that can't be the only reason. At my job we are free to choose. Some have MacOS, almost everyone else has Windows, I am the only one using linux. But the joke is that Windows users are heavy users of WSL and RStudio. We are not forced, or even suggested to use Teams, Office or whatever tooling Microsoft has, and in fact meetings were I am invited had to be done in Zoom instead of Teams only because of me.
I believe that it is more a familiarity issue, and seeing Linux/Mac as an obscure tool only usable by gurus.
I think your experience is the exception rather than the rule, most people work for a company that dictate the computer and OS to use, as well as the communication tools (teams, slack...) for which they have an enterprise license.
My comment was more on the line "when given options, people will massively choose Windows" just out of familiarity or fear of the unknown, and by people I also mean management. If I was manager, I wouldn't force Linux (which I use exclusively since c. 2000) unto my workers, unless I am ready to set Linux familiarity as a CV minimum requirement, reducing my hiring pool by 90% at least.
The parent comment implies there is some dark conspirancy behind the scenes to force Windows unto us. There might be a couple decades ago, today it is just what people choose if given choice. Windows 11 nags its users all the time, yet they accept it as it was normal and unavoidable.
It's not so much the AI vendors that are the issue. It's everyone else that want to grab some of those LLM profits as well. Apple in iOS/macOS, Google and Samsung in Android, Microsoft in Windows/Office, Raycast, Meta in WhatsApp, a company's helpdesk, random websites, etc. Everyone wants to make a profit by selling you an AI subscription or by reducing the number of humans employed.
> The choice was made from someone else, and you just accepted it as normal
That behaviour in itself is believed to be morally wrong, and dystopian. That still doesn't answer the question why people just accept this. How can "we" deem ourself more moral than the Nazis, and literally do this.
The statement "human civilization lives on", as far as capitalistic interest is concerned, stands true even if the ones who will survive are ultimately the oligarchs, and all art form is machine generated.
In a discussion about cost effectiveness, how subjectively fun the writing feels like to the reader isn't a factor, except maybe if we were working on writing comedy.
>Why have we not seen an improvements in products?
This was never the goal, because it would influence the standards and expectations of the end user. Any gamer remembers when Baldur's Gate 3 came out everyone said somewhere between "this is the new standard in AAA games" or "this is an amazing value for money". The response from the rest of the gaming industry was pretty much "lol no, not really".
Customer satisfaction rarely aligns with C-suite goals. The faster you understand this the better for your mental health.
If customer satisfaction actually aligned with corporate goals then planned obsolescence wouldn't be a thing. Yes, this doesn't apply to software... or does it?
this overly cynical take is bad and wrong. c-suite does in fact care about customer satisfaction; it's why you get asked about it in surveys all the time. it's why a game will live or die by the review scores.
while it is true that c-suite behavior does not always align with improving customer satisfaction, that's usually because c-suite is dumb and bad at their jobs.
the enshittification we experience on social media, on the other hand, is due to the fact that we are not their customers. we are the product.
The vast majority of the Yu-Gi-Oh Master Duel community has been begging Konami through surveys to remove cards like Maxx "C" from Master Duel's best of 1 format because it's the card that established Master Duel as the "coin flip format".
Guess how much Konami cared. Just because there's a survey doesn't mean that they're looking for any kind of feedback, but for very specific feedback. It's not Konami's balance team that tried to make for a fair game, but the C-suits that want to keep printing cards and keep power creeping, because the target isn't fairness or balance, but it's keeping the players who want to stay competitive swipe their credit cards more often.
reply