The contracts are really worthless if for example Trump wants to share the transcripts with Putin. At this point I don’t understand why governments would outsource any of their data services to 3rd parties (I.e outside of the EU). EU should establish an EU gov cloud program and just run the models themselves for all the EU countries(under license of course). They would probably get better bargaining power as well.
You’ve leapt to the most extreme scenario to declare these contracts worthless.
Of course something like what you describe could happen. But under the central case, ensuring some privacy (with the ability to seek damages if breached) whilst still providing access to frontier models seems like a reasonable trade-off with a lot of value.
When it comes to government data, contracts are often much stronger than you may be used to in the private sector, as mishandling the data can be criminal in nature. And data sovereignty requirements are common. This stuff is table stakes when selling to governments.
Such a thing would require building data centers, i.e. industrial compute factories, but the EU is politically paralyzed from doing that because electricity is already outrageously expensive and there is just sort of a general popular outrage about the prospect of it.
> the EU is politically paralyzed from doing that because electricity is already outrageously expensive and there is just sort of a general popular outrage about the prospect of it.
That is called democracy. Where the good of citizens is valued above what big corporations want. To push up pollution and power prices so big American tech companies can play with AI seems something that only a mad king would do.
If AI can find cancers faster than a human, which they've already proven, it's tradeoff that's worth it. It's only getting better and there's no telling how else it will benefit humanity.
Governments are always at least a decade behind when it comes to technology. Just don't complain when they have massive data breaches weekly, because they decided their principals were more important.
Your latter point belies the former: The the trade-off is only worth it if the data centers are used for purposes that benefit humanity. As yet, all we've seen them used for is wealth consolidation (through labor force reduction), mass surveillance, hacking, and other antisocial behaviors. There's no reason to believe those won't persist as the profit-maximizing use cases.
It's not democracy. People don't want high energy prices. It's just their only voting options are all politicians who will give them high energy prices.
Big corporations are only big because they provide what customers want. They don't get to extract money from people's work, the way politicians can.
Germany spent most of the last two decades shutting down nuclear power plants while blocking renewables because the politicians were sponsored by coal and gas companies.
It's also because they were addicted to Russian gas, and had to spend a fortune to wean themselves off it once the war was raging - though not before giving Russia a lot of gas money for the war - and converting its power plants to use LNG, which it ships in from the USA. That is a mad situation even before Iran.
Citizens are dumb though, the stupidity of the public killed nuclear. Not every decision needs to be democratic. Certainly not core infrastructure decisions.
It's also so easy to astroturf to manipulate public opinion. If I were China, I'd post so much anti-datacenter content everywhere in the West to gain a leg up in the race to superintelligence, because we are dumb enough to let our citizenry influence these decisions.
Imagine if we did the same thing for the nuclear race. ASI is winner-take-all, the first one to ASI can just lock everyone else out.
The recent growth of LLMs is a black swan. Any future prediction you attempt to make has massive error bars.
Europe is in a very precarious position. If LLMs actually play out and lead to economic booms (I'm not saying for individuals, note that), the EU as a whole will get absolutely trounced economically and likely experience a shrinking economy for the next 20+ years along with a shrinking population in most of the member countries. They are so far behind on power and compute of the US and China there is not any current path to catching up.
The US itself is in a precarious position with the amount of debt it has accrued, it could be right that LLMs are what leads to future economic booms, but be so far in the hole it takes decades to recover.
China is really in the best place at the moment. Trying to build up power and compute as fast as it can, copying American innovation at the cheap, and having a mountain of US and EU money to spend on it.
So our technoverlords should decide for us? All these arguments to go back to the middle ages and feudal systems are based on the idea that we should be oppressed and our opinion should not count. That thinking destroys enlighten societies and bring inequality and poverty.
It would be a mistake to blindly assume that democracy can survive the technical future. The rise of democracy went hand in hand with the importance of skilled labor. If the need for labor dies, at least in my eyes, democracy dies with it.
If I was a American oligarch who became president due to my ties to social media platforms and influence I would also advocate european governments use American AI services.
AI is local. They need to invest in procuring hardware, which means making it themselves if that is the only way to get a card in every computer that can do the job.
If cloud based AI is any good, they will sell versions that can be hosted locally just like every other service since the beginning of the internet.
I think one of the AI 2027 videos covered this pretty well. EU cannot make nor really afford the compute they need to buy. All they really have the ability to control is UV machines for making masks which is going to get politically hot very soon.
The US and China have claimed all production for years to come.
How is electricity outrageously expensive? We get power in the most flexible of forms (electrical) delivered directly to our homes at a rate that no human in history would have ever performed voluntary physical labor for (the equivalent of thousands of horses or people for a few hundred bucks an hour, assuming <500$/MWh). Even complete energy autarky has become feasible for private individuals if you are willing to invest.
Just because power used to be even cheaper in the recent past, by doing non-sustainable shit and pushing all the resulting costs onto others/the future does not mean that was ever appropriate, necessary or desirable, and everyone advocating for chasing that model is a greedy, selfish asshole in my view.
What's reactivity in this context? How quickly a website responds? My personal experience is that fast backends with server side rendered HTML and a bit of dynamic sections are in practice MUCH faster than SPAs, 20 years after using Gmail for the first time. Gmail then was fast, 95% of modern SPAs are slow.
"State management" is a fancy way to say "your Javascript stores data in variables in some way, which are updated as and when needed".
"Reactivity" means "when those variables change, some generic code already exists to update the DOM instantly". As opposed to writing specific pieces of code to make those changes directly.
Unfortunately, with React, "updating the DOM automatically" generally means: you update the data; React recreates the entire DOM as a new copy in memory; React runs a diff checker to find out what changed from the real DOM; React makes each individual change in the real DOM. With React (or other virtual-DOM-centered alternatives) being used for the smallest of web projects, so many small sites do all that work when the procedure ought to be as simple as: you update the data; you update the DOM directly.
It comes down to inexperienced devs not being able to consider which tool is best for the job at hand, and just hammering nails into every web project in the same way.
I mean all you have to do is change the framework to a better one. Vue with Vapor mode, Svelte, Solid or Preact. These don't have a VDOM at all or use a super fast minimal one.
Signals are basically as fast as writing event handlers yourself but way less fragile and verbose. React is just shaping everyone's impression of all JS frameworks because it's very popular and very slow.
Reactivity is updating the DOM whenever the underlying data is updated. That's one of the most basic reason for existence of all js frameworks. There's TC39 Signals proposal to bring that natively, but for now as said above the choice is either use a framework or recreate one. It's surprisingly easy nowadays using Proxy.
It’s always been easy to reactively update the DOM. I was doing that in vanilla JS before HTML5 had been formalised.
What the frameworks do is take this concept one step further to create a SPA (single page application) but I honestly think they’re an antipattern that needs to die.
The problems are ownership and provenance mainly, and that all DOM access is slow and requires syncing code internally.
Anyone can write code that tweaks nodes or uses templating, but doing it in a consistent way on a project with hundreds of thousands linea of code is challenging, especially if you also need performance and everyone's least favourite child: accessibility.
Sure you don't get some of this with frameworks, but they help a lot.
Well yeah, obviously frameworks are intended to provide a standard interface for projects that scale. But you’re describing a different problem to the one originally argued.
> Reactivity is updating the DOM whenever the underlying data is updated. That's one of the most basic reason for existence of all js frameworks.
Most web apps and websites do not have that much state variables (even though it may be a lot of data). So while React may offer a small dev speed boost, it’s consumed by the need to build stuff and a whole deploy pipeline to nurse.
Libraries like HTMX and Alpine can take care of most the boilerplate for old style imperative patterns. And backend frameworks like Rails, Phoenix, Laravel donhave their own implementation for data updates.
I click the upvote button and a number goes from 0 to 1. Somewhere else on the page a banner gets displayed saying you have upvotes.
Your javascript can target those 2 divs, ala doc.getElementById(counter).innerHtml and document.createElement.
Like that kind of works for simple stuff. But it gets more complicated fast. Reactivity is when your counter listens for change in the count and updates itself whenever that count changes.
That's why you wouldn't target those 2 divs directly, but have one dedicated place, in which you update the divs and call that place from any other place that needs to update the 2 divs.
Well, that's simplifying the situation, but as those situations become more and more in ones site, at some point one might switch to a mediator pattern, which kind of is very close to signal emit and listening for signals.
Beautifully ironic. This post argues against being convinced by LLM statistics, by showing (presumably) human statistics with so many "Don't Trust Me" flags.
1. Unverifiable source ("a meta-analysis", trying to get authority without actually having it.
2. Hyper-specific numbers without error bars (and source), pretending to have certainty and precision.
3. Shockingly large number, intended to get an emotional reaction and a sense that it rounds to "all".
4. Numbers align too neatly with the point being made. They could still be true, but there could also be some overinterpreting, stretching of the truth or just plain lying going on.
Beautiful demonstration that numbers intended to convince are never to be trusted outright. No notes.
As an aside, the whole freak-out about LLM hallucinations has mostly shown me that way too many people accept human numbers way too easily. People are frequently as dumb or even dumber than LLMs.
One of my software engineering maxims is that popularity is a feature.
I don't mean that in a "social proof" kind of way. Popularity brings it own advantages. Because lots of people use something, you get the advantages of lots of other people using it:
a larger ecosystem, more libraries, easier to find new team mates, easier to find people willing to learn, better and more answers on Stack Overflow (or in LLMs now, I guess), etc., etc.
And there's just no compiled languages more used and known than JavaScript/TypeScript. Java and C# come closest, but they're still far off and still require runtimes.
I've been using LED lights for over a decade. Can't say I have replaced any of them.
And nowadays you can get E27 LED bulbs for 75 cents at IKEA, which is cheaper than incandescents ever were. Not sure how they managed that, because it sounds insane.
Similar experience. I'm perplexed at the people saying LED bulbs fail frequently. I have probably 100 of them in service for 10 years or so (basically since Costco began selling them) and possibly one has failed. Memory is hazy but definitely no more than one.
No, I have the same problem with pockets. Ear buds love falling out of my pockets in the car.
It's the main reason I mostly wear jeans. The five pocket configuration's vertical orientation holds onto its contents better than the slanted pockets of slacks and chinos.
I would really like some slacks with vertical deep pockets. Jeans get so boring.
Right. My earlier comments are not aimed at jeans, they're pretty much the exception. Their two main pockets being positioned at the front make it difficult for objects to fall out unnoticed.
I'm guessing, but I'd imagine as jeans first became popular in the days of horse riding side pockets really weren't a practical option, the rider's motion would contribute to things sliding out of them.
My complaint is that modern clothes designers have forgotten or stopped implementing good pocket design that was fully solved several centuries ago.
"I would really like some slacks with vertical deep pockets."
So would I, and deep pockets do exist but unfortunately not in mass produced pants that are commonly available (one's regular clothing). I have some army BD fatigues/dungarees with really good side pockets. They're not only deep but they're cut in a sort of a shallow 'P' shape. The P-cut at the bottom forces keys etc. to slide down one's side at 90° to the pocket exit when one is sitting. It's thus almost impossible to accidentally lose something out of them.
The questions are why don't manufacturers just copy this excellent design and why don't more people complain. If people stopped buying pants with shallow pockets the problem would be solved overnight.
The result is the only thing about exercise that I enjoy. It's the data, not the feelings. I don't like going to the gym. I don't like doing my exercises. I don't feel better afterwards (usually the opposite: I'm tired and sore). I don't compete or compare myself to anyone else. But I do like seeing that e.g. my bench press is 5 lbs better this week. That's the entire reward for me.
Yeah, and probably also forcing Zero Data Retention contracts on those service, instead of the consumer variants.
reply