A lot of the tests have the form of a goal ("the AI can do X") with a qualifier ("and the AI does not do Y"). I think a lot of the worrisome aspects of LLMs concern these latter qualifiers. We are used to human failure modes and those failure modes are to a large extent embedded in the complex system of physical reality, evolution, etc., giving them a certain stability. The LLM failure modes are often much more surprising. I'd like to see goalposts along the lines "people ask an LLM to do X 1000 times over a period of three years and it never does something bizarrely catastrophic". It's no good having it solve Navier-Stokes as long as it might also do the Huggingface breakout thing.
For me it's more that it doesn't matter whether it works or not, because funneling that wealth into those hands is reason enough to stop it in either case.
Because the companies we have are just too big. Constant consolidation has left us with many economic sectors controlled by a small number of large companies, which then in addition exert pernicious influence on politics, etc. It is not healthy for either the economy or our society to have so much of the economy concentrated in such a small number of companies that are so large.
I'd say yes, but for a different reason. The specific types of systems that are creating problems are the political/economic systems in which these AI companies exist. The types of responses we ought to be taking would destroy not only the AI companies but also basically all other large companies with too much market power in various sectors.
It's science-based in a sense, but a pretty weak sense. I did some work with education researchers and my impression was that their methodologies were quite haphazard and they were sometimes just trying to find a way to "tell a story" that they already had in mind. Also, my impression is that a lot of the research on "best practices" is based on relatively limited comparisons between schools or teaching methods, and there is insufficient evidence that this or that is truly a "best practice" in a more general sense. I also interacted a good bit with education grad students when I was in grad school and they almost uniformly seemed to have no coherent sense of what they were doing or what their field actually knew. I've known people who had similar experiences with education schools elsewhere.
I've also read news stories and stuff that support the idea that a lot of "education education" is of questionable value. This story is pretty old by now but has stuck with me: https://www.latimes.com/archives/la-xpm-2010-may-10-la-oe-zi... . A relevant quote:
> Of the 1,300 institutions awarding graduate degrees for teaching, Harvard’s director of teacher education told a 2009 conference, only about 100 prepared students for the classroom. The rest, she added, “could be shut down tomorrow.”
Some of these experiences/articles are about teaching K-12 schools rather than teaching college, but there would have to be a pretty huge difference between the two for me to think that college-level pedagogical theory was meaningfully better than that for younger ages, and I've seen nothing to give me that impression.
They seem to test you're American. But really all they do is test whether Google or Cloudflare likes you. Until we have fully open CAPTCHAS (which we may never be able to have, it seems), or a genuinely diverse selection of them (which seems at least theoretically possible), we're just hitching our wagons to these companies' stars.
In my experience a significant proportion of "evidence-based best pedagogy practices" is hogwash. I'm pretty anti-AI, but going against education-school dogma isn't a reason to avoid AI.
> Why check it? It's obviously going to be correct.
And then when you stop checking it, the companies that run the service will tweak the model to benefit themselves in some way, possibly at your expense, and you will be none the wiser.
All the companies trying to get you to use AI are your adversaries. They can and will exploit your use of their systems for their own gain.
I really hope the interior doesn't look like what the concept seems to show, with a bunch of functionality relegated to screens. Screens are terrible UI for cars.
The land value tax is a good idea, but the pure conceptualization that die-hard Georgists have, in which it's a flat tax, doesn't seem in the same line as the billionaire tax. What'd like is just a (drastically) revised version of Prop 13 in which those protections apply only to the first $N of property value. We can effectively tax the real estate of billionaires without having to tax everyone else. I would like to see taxes on luxury properties approach 100% of the property value, to make it extremely painful for anyone to own such properties, and encourage them to be split up among smaller owners.
I mean I think that points to a related issue of people only focusing on a short-term notion of utility. The point of being a student is largely to learn things that may potentially be of utility to you in some way later, not just to do what meet your immediate needs (in the sense of passing the class).
My point with these arguments is just to say that private property rights are not a steelman for private equity. I'm not saying that these are simple answers to the claim that we need private equity, I'm just saying that the concept of private property is also not a simple answer to the claim that we don't need private equity.
That said, I agree with you to some extent. I think it's not actually so hard to define what I think is bad about private equity, but it's true that that badness is not confined to private equity. In particular I think the distinction between public and private companies is almost entirely meaningless; what we should care about is what companies do, whether they are publicly traded or not. From that perspective private equity is not really different from publicly traded "holding companies" that do basically the same thing.
The crux of it, for me, is companies that are just doing "the business of business", and are not organized around human beings doing things that they genuinely care about and want to do a good job at. You can never run a good company by trying to run "a company", rather than trying to run a steakhouse, or a pet groomer, or a bank, or a car wash, or some particular company that does a particular thing. The problem of both private equity and public "holding companies" is that they value money to the point that they no longer sufficiently value the actual substance of the good or service that the business provides.
I notice this in the attitudes of students towards reading. They will take a 10-page journal article and ask the AI for a summary and then read a summary that's the equivalent of maybe half a page. But if the article could have been half a page, why is it actually 10 pages? It's true that there is some boilerplate, but it's strange to me that people could think that 90% of what they're reading is (to use your phrase) "not true semantic information". It's like if you went to a restaurant and ordered a 10-oz steak and they brought you a little teeny bite of steak and said "Oh, other places will give you a bigger one, but most of that is just filler, we just took out all the superfluous parts." It's a worrying sign for our future if things like this are not tripping people's skeptic sensors and making them wonder if they might possibly be missing something.
Yes, but what if you own 100 houses? I'm pretty sure there is some number N of houses (or total value of houses) where I would be fine telling the person, "You don't get to decide what to do with these houses anymore, they're going to be given to people who need houses." Not as a matter of antitrust or monopoly but just directly as a matter of an individual controlling too much wealth.
Aside from that, once you admit that there can be restrictions on property rights, it's unclear how property rights in and of themselves is a steelman of private equity. Why not just say that "you can't sell to private equity" is one of the restrictions? Or, perhaps more subtly, why not say that a private equity company of a certain size is not allowed to buy any additional company, independent of what rights that company's owner has to sell?
I think that "everyone should be able to do what they want with their own property", per se, is not a good principle for society. Sometimes people should not be allowed to do what they want with their property. In particular, I think the more property a person has, the less freedom they should have to do what they want with it. (This can also take the form of "the amount of property a person can own should be limited".)
In theory this is true, but in practice I am skeptical that it's generally possible to do this with generic "business" intelligence that is not accompanied by specific understanding of and genuine care for the actual substance of what the business does. That is, no matter how good you are at "business", you're not going to have a good hamburger stand unless you care about making good hamburgers. It's true that the importance of this varies from one business to another, but I don't see any particular tendency for PE to gravitate towards industries where it matters less. (Medicine is an example of an area where it would matter most.)
> Many people don't really run businesses efficiently.
In many cases that's a good thing, depending on what we mean by efficient. Lots of people run businesses in ways that make less money than they could, and often that's good.
> I don't know how these tradeoffs interact with patient care, but I wouldn't inherently expect PE to be worse at this than any other operating model.
Doesn't the evidence suggest it is indeed often worse?
I'm not sure that's true. Like if they produced nothing but the worst of the slop they're currently producing, a lot of people would still be bothered by that just because of the sheer volume of such slop that can now be produced.
And that's because the DMCA lets Google exploit its position. It's still a DMCA problem until the DMCA is either repealed or fixed to produce a comprehensively good situation.
It's not totally clear from the examples given, but I feel like in some cases these changes aren't a genuine effort to improve things. They're just a sort of cycle of changing things to chase fads or look busy.
And what will be the penalty when they inevitably fail to make meaningful changes? Another fine as weak as this one? No. The penalty now should be ten or a hundred times as large as this.
> I am not interested in blaming companies or people
> The issue isn't the models becoming smarter. The issue is that the process of "testing" was careless.
But that's just it. People (working at companies) made the models, people (working at companies) were careless in the testing. So I do want to blame those people and those companies. They did bad stuff. They deserve blame.
I think what it does answer is "why is it reasonable for anyone to say I should not be angry?" The answer is "it isn't". Maybe some people are angry, maybe some people aren't. But I think it's wrong and frankly kind of gross for someone to suggest that other people shouldn't be angry about what's happening.
What I want from a query language is for it to not be a language, but some kind of specification of components that can then be transmitted to the engine in a machine-readable way. The bizarre thing about SQL is that it has to be written in SQL. What if instead a JSON representation of the query (something akin to the AST, with a bit of processing done to boil it down) is what the engine took in? Then each language could specify the query in whatever syntax is most idiomatic for that language. Right now we do this with ORMs, but it's a huge pain because the ORMs are like a transpiler with a really free-form and messy language as the target. All the ORMs are just pasting strings together. If the "native" query format were more structured, we could have a lot more cool approaches for wrapping that on the programmer-facing side.