The bright side is that it should eventually be technically feasible to create much more powerful and effective guardrails around neural nets. At the end of the day, we have full access to the machine running the code, whereas we can't exactly go around sticking electrodes into everyone's brains, and even "just" constant monitoring is prohibitively expensive for most human work. The bad news is that we might be decades away from an understanding of how to create useful guardrails around AI, and AI is doing stuff now.
I get the feeling that you're winging the specific numbers because they're spectacularly incoherent.
But anyway, the United States is extremely rich and has essentially no big problems that can be solved by a small amount (say, a few billion) of money. The problems are either so big that it would take trillions to solve (supporting aging population etc), or blocked by something other than money (politics, regulations, etc). The big problems that can be solved just by throwing a few billion at them are solved quite easily by either the government or by private entities like the Gates Foundation.
In practice, it seems that politics generally takes precedence over problem solving. If you look into the psychology of it, neither politicians nor voters are really incentivized to solve big problems. This is especially true for big problems that will take more than an election cycle to solve.
It seems to me that it would be easy to support an argument that suggests more big problems could be solved if incentives were better aligned toward problem solving and if competent people, not professional politicians, were chosen to solve them.
Even the US median household income is "only" $83k. Looking at stuff like this + the rest of the blog I'm not convinced the author is any less out of touch than the people this post is criticizing.
I think being in debt is too normalized. I know there are specific financial instruments, that if one is savvy, can use to be in debt but come out ahead. But for the average person, I do not think mortgages and loans should be so widespread. Schools don't teach the upper level of financial suave, and neither do most parents who aren't already wealthy.
We should normalize buying a home and owning it outright, even if it takes more up-front capital. It's something that should have been done when housing was cheaper, if you ask me. Me and my girlfriend watch a lot of House Hunters on HGTV and almost every episode concerns someone buying a house with the expectation of mortgaging, so that they can get a house beyond their immediate means.
They are issued for something [contingently] incredibly valuable that cannot be repossessed. They are also generally issued at a lower rate of interest with little collateral, in part due to the fact that they cannot be discharged in bankruptcy and the resulting lower risk.
[edit: added "contingently" above. Some education programs have been found to be scams yet could be paid for with debt. I generally stand by the value of higher education and have found it to be a net benefit in my own life. The value has been strained as costs have shifted and the social contract is rearranged behind the backs of educators at all levels]
There's nothing rigged about loans being made for something that can't be repossessed and without collateral, there's a whole segment of the lending market for that (consumer lending, eg people take out loans to go on vacation). So absolutely nothing rigged about that. And those loans can still be discharged in bankruptcy. There's a government guarantee, but that's at least as much an advantage for the lender as for the borrower. The relatively low interest rate is a result of that, the loans aren't high risk for the underwriters. Government guarantees aren't unique to student loans either btw. The only thing that's really unique is that the loan can't be discharged, and that is very much a disadvantage for the borrower. So student loans are rigged against the students. No bank would want to make those loans under normal circumstances, so the government has rigged the game in their favor to entice them into the market.
Education is a unique product, which I'd argue warrants favorable loan terms. Ignoring the inability to declare bankruptcy.
Even in non-scam programs, there's sometimes a massive amount of unknowable hazard on the path to a degree. And whether or not the degree is attained, the debt is still due.
In terms of hazard, I don't mean grades. Assume perfect grades. Especially in graduate programs, interpersonal politics can permanently wreck students who then have no effective recourse.
Examples and intricacies of what I'm trying to imply, aside, consider that education is a unique product that costs a lot of money but, in the process of attaining the product, you aren't treated as a customer but as if your position is precarious. Grades aside. Whatever specifics you might imagine, as they would vary, assume that the system is set up to protect the school that is collecting money. Not the person paying it.
Coming back around to my ultimate point about loan terms and risk.
I don't know how intentional your choice of words was, but describing education as a 'product' is a sad reminder of the current state of the world. It's as if every action and experience in life must be translated into money.
Colleges, let alone Universities, used to be much fewer in number.
It used to be more widely known that the better Colleges, that is mostly the Ivy League and the top Liberal Arts colleges, were the only educational institutions that mattered in terms of imparting the vaguer yet more elite type of education of which you are likely thinking.
That is, the institutions who reliably facilitate the fuzzy process of becoming an educated person who can think like one. In the elite sense of these things.
What we've lost to an extent, in the multi-decade long taxpayer funded college degree bonanza, is that it is these elite institutions which are still seen as the only colleges that matter in the sense that I think that you are implying.
Other colleges are not truly valuable in this sense. So what are they good for? They are good for work training and certification pathways.
Universities have no problem charging obscene rates, which is hard cash, for something that they too would like to be only distastefully associated with money when debating the merits.
If it isn't translatable to money, then make it cheaper. Or keep the value firm and high, but also let's talk about what is being sold.
The 'value' of the education can vary widely. Both in terms of value meaning the ability to use it as a credential for high-paying jobs, or in terms of the imparted skillset/knowledge that either make you a better person or enable you to achieve things like starting a business/inventing/etc.
I mean there is a HUGE spectrum here. And this means that there are many educations paid for with debt that are utterly useless (and therefore valueless) on both of these metrics. In other cases where an investment failed to this extent, you can declare bankruptcy. But not student loans! No no no. Those are forever. Hardly a favorable condition for the borrower.
I'm curious too. The rates aren't particularly generous. Okay, maybe they're a bit lower than the private markets might deliver given the rates of default. And maybe they let you study whatever you want. A private lender might insist on funding economically worthwhile degrees.
The real issue isn't the loan, but the entire system. The colleges have really jacked up the price of a degree. That's the real source of the problem. The interest rates are much closer to the market rates for capital.
I don't know that I'd use the word generous but they do have to advantages relative to other types of loans:
1. A relatively low rate for no collateral. It's never the 2.x% we got on mortgages during the brief window when mortgage rates are good, but compared to other rates you get showing up at a bank with no collateral, they're amazing.
2. Almost guaranteed approval for the above. Can you imagine an 18 year old walking into a bank with no credit history and collateral and walking out with a loan for hundreds of thousands of dollars? Sure, it may be a terrible idea, but a student loan is one of the easiest to get.
But it's for life right? You just can't bankrupt out of repayment, and since its in the US, missing a payment kill your credit score and prevent you from borrowing for anything.
In the 80s-90s in my country, we imported us-style loans, 'revolving credit', that were later called 'credits revolvers' because once you've taken one, a revolver was the only way out (not a joke, I know of 3 people who died in my village from this). We had to allow personal bankruptcy, and once the law was implemented and used, interest rates on those kind of loans doubled, and now they don't exist (although klarna and co are trying).
It's frustrating how far you can go out of your way to avoid being associated with such superficially similar tropes and still fail miserably. Yudkowsky in particular hated that he couldn't get a discussion without being typecast as the guy worried about Terminator. He hated it to the point he wrote a whole article on why he thought Terminator tropes were bad (https://www.lesswrong.com/posts/rHBdcHGLJ7KvLJQPk/the-logica...).
As a side note:
> any serious attempt at discussion gets bogged down by [...] without taking a single shower in the same span of time.
This is unnecessary and (somewhat ironically) undermines your own point. I would like to see less of this on HN.
> I'd argue that the biggest reason machines are black boxes are because no one is bothering to look inside of them.
People do look, but it's extremely hard. Take a look at how hard the mechanistic interpretability people have to work for even small insights. Neel Nanda[1] has some very nice writeups if you haven't already seen them.
> Very few are trying to understand why things are working
What is in question is why this is given so little attention. You can hear Neel talk about this himself. It is the reason he is trying to rally people and get more into Mech Interp. Which frankly, this side of ML is as old as ML itself.
Personal, I believe that if you aren't trying to interpret results and ask the why then you're not actually doing science. Which is fine. There's plenty of good things that come from outside science. I just think it's weird to call something science if you aren't going to do hypothesis testing and finding out why things are the way they are
The problem is that mechanistic interpretability is a lot like neuroscience or molecular biology, i.e. you're trying to understand huge complexity from relatively crude point measurements (no offense intended to neuroscientists and biologists). But AI wants publishable results yesterday. I often wonder whether the current AI systems will stay around long enough for anyone to remain interested in understanding why they ever worked.
People will always be interested in why things work. At least one will as long as I'm alive, but I really don't think I'm that special. Wondering why things are the way they are is really at the core of science. Sure, there are plenty of physicists, mathematicians, neuroscientists, biologists, and others who just want answers, but this is a very narrow part of science.
I would really encourage others to read works that go through the history of the topic they are studying. If you're interested in quantum mechanics, the one I'd recommend is "The Quantum Physicists" by William Cropper[0]. It won't replace Griffiths[1] but it is a good addition.
The reason that getting information like this is VERY helpful is that it teaches you how to solve problems and actually go into the unknown. It is easy to learn things from a book because someone is there telling you all the answers, but texts like these instead put yourself in the shoes of the people in those times, and focus on seeing what and why certain questions are being asked. This is the hard thing when you're at the "end". When you can't just read new knowledge from a book, because there is no one that knows! Or the issue Thomas Wolf describes here[2] and why he struggled.
Error correction is possible even if the error correction is itself noisy. The error does not need to accumulate, it can be made as small as you like at the cost of some efficiency. This is not a new problem, the relevant theorems are incredibly robust and have been known for decades.
Can you link me to a proof demonstrating that the error can be made arbitrarily small? (Or at least a precise statement of the theorem you have in mind.) I would think that if the last step of error correction turns a correct intermediate result into an incorrect final result with probability p, that puts a lower bound of p on the overall error rate.
> Math fans tend to discount and dismiss applied statistics as being not math, in a way that they don't do for physics, for some reason I don't fully grasp.
It's because we're secretly afraid that the physicists are smarter than us.
Less facetiously, physicists keep discovering things that lead to new mathematics we would never have dreamed of ourselves, so we have a healthy respect for how insightful they can be.
