Has Europe given up on tech?

[lfromnj]

That’s a common dynamic on the continent—and in the United Kingdom. The U.K. is hosting a summit on AI safety this year, and British Prime Minister Rishi Sunak wants the country to be a world leader in “the governance of AI.” Of course, few of the most important AI firms—except Google’s DeepMind—are located in the U.K. The U.K. has little to no chance to lead in AI development, but it is eager to lead in AI regulation.

That’s a common dynamic on the continent—and in the United Kingdom. The U.K. is hosting a summit on AI safety this year, and British Prime Minister Rishi Sunak wants the country to be a world leader in “the governance of AI.” Of course, few of the most important AI firms—except Google’s DeepMind—are located in the U.K. The U.K. has little to no chance to lead in AI development, but it is eager to lead in AI regulation.

The optimistic outlook is that Europe is simply overzealous in an understandable urge to regulate new technologies, and that poorly written laws will be refined over time in both practice and legislation. The more cynical, pessimistic outlook is that Europe has simply given up trying to win in tech and now only seeks to regulate and fine the United States’ tech giants out of a sense of ressentiment. After all, who do these EU bills largely target? The EU recently designated 22 “core platform services” under the DMA to be subject to strict oversight. Twenty-one of those services are American, with Chinese-owned TikTok as the sole exception.

https://foreignpolicy.com/2023/10/23/metaverse-europe-uk-us-big-tech-regulation-innovation/

Yuropoors are poor and salty they don't have any innovative companies so they decide to go into regulative protectionism to collect fines from our wonderful American companies.

[Okay, maybe Mike Johnson is a competent parliamentarian.]

Absolutely deranged technoutopian framing. Europe hasn't so much "given up on tech" in that "tech" is an obviously good thing that it just doesn't want to do as it has "given up on tech" in that its governments and bureaucracies have decided that "tech" is something whose proper role in an economy and a society should be soberly and prudently assessed just like everything else. If European policymakers are triggering people who think that "tech" is some sort of killer app for human suffering as badly as they apparently are, then they must be doing something right, even though a lot of these regulations do seem badly written or badly implemented.

[Vosem]

Absolutely deranged technoutopian framing. Europe hasn't so much "given up on tech" in that "tech" is an obviously good thing that it just doesn't want to do as it has "given up on tech" in that its governments and bureaucracies have decided that "tech" is something whose proper role in an economy and a society should be soberly and prudently assessed just like everything else. If European policymakers are triggering people who think that "tech" is some sort of killer app for human suffering as badly as they apparently are, then they must be doing something right, even though a lot of these regulations do seem badly written or badly implemented.

I think in the ultra-broad framing of "tech" where it means 'new ideas and economic paradigms' it really is an obviously good thing, and a society which thinks that new ideas and economic paradigms should be judged by impersonal bureaucracies rather than by ordinary people in the marketplace really is doomed to a dreary stagnation, where new ideas will never challenge entrenched power structures.

(Not that I'm really absolving the United States here, which created the NRC in order to slow the growth of nuclear power in the country, and which created the FAA more-or-less with the purpose of making personal aircraft unaffordable for the upper-middle-class, but this problem seems far worse in Europe, where certain countries -- for whatever reason particularly Italy -- have responded to every advance in informatics or medicine over the past decade by feverishly restricting it and trying to make it is inaccessible as possible for ordinary people.)

It's hard not to see the numbers lfromnj is showing, which depict an educated and developed society (really, many educated and developed societies), as not a horrific indictment of that society, and a sign that something has really gone horribly wrong, and probably in a way that applies beyond "tech".

[Хahar 🤔]

I got placed on a new project at work today where my job is to help determine the scope and revenue impact of changes that will have to be made as a result of the EU enacting the Digital Markets Act, so thanks to the European Union for ensuring my continued employment.

Obviously most of you know where I come from (I grew up here, I live here now, most everyone I've known all my life has been employed in the tech industry, and I am currently employed in the tech industry), but, with that disclaimer, I would say that just about everyone I could possibly talk to who would have an opinion on this subject would agree that European Union regulations are usually vexatious and often born of a lack of subject area knowledge. The article that lfromnj has linked provides plenty of examples of this (European data protectionism policy is particularly bizarre and arbitrary), and even Nathan does not deny that this is the case, even though for reasons that I don't really understand he seems to think that it is a good thing.

The article states the reason that this is the case, but it would be obvious even if it hadn't: providers of technological services are generally not based in the European Union, and so capricious and arbitrary EU regulations do not hurt EU companies. If anything, to the extent that policies like data protectionism compel companies to maintain data centers within the EU, they provide employment and serve as a benefit. The costs of these policies tend to be borne elsewhere, by American companies that have to spend money on projects like what I'm working on now and by consumers everywhere who deal with worse products (charging ports that are less efficient, online reviews that are censored) to meet the standards of the European Union. It's an easy way for eurocrats to claim a victory.

My view is that in general, rather than forcing everyone else to deal with EU legislation, it would be preferable to sever the European Union from the rest of the Internet, so that we can enjoy functional online services and Europeans can enjoy the right to be forgotten. Ideally this would lead Europeans to complain to their elected officials and get better legislation, although in consideration of the way that the European Union works this would probably not accomplish anything. To some extent this does happen already; plenty of people have noticed with frustration that once you add an EU resident to a Facebook Messenger group chat it is permanently impossible to create polls within the chat. Fortunately Europe is not so large a market that it can't just be ignored.

[Aurelius2]

Rishi Sunak is the only national leader taking the dangers of AI seriously. This is to be commended and does not deserve to be lumped in with the continental Europe loony bin of bad ideas like using state power to censor what the state deems "misinformation", making us all suffer from endless popups about GDPR and cookies, and trying to ban people from emailing their boss at times the state disapproves of.

AI development is hopelessly entangled with a deranged suicide cult. Half of this cult looks forward to the day in which, as they see it, we will all be "liberated" from the "tyranny" of physical existence and all be severed from our physical bodies, to be turned into bits in a machine where AI is constantly feeding our (simulated, digital) dopamine receptors with fentanyl-potency hedonium. The other half looks at the concerns about existential risk and says, Well Actually that's a good thing, and the sooner it exterminates us to turn the planet into computronium the better.

The Butlerian Jihad shall be waged against god-like AI, not browser cookies.

[Okay, maybe Mike Johnson is a competent parliamentarian.]

Obviously most of you know where I come from (I grew up here, I live here now, most everyone I've known all my life has been employed in the tech industry, and I am currently employed in the tech industry), but, with that disclaimer, I would say that just about everyone I could possibly talk to who would have an opinion on this subject would agree that European Union regulations are usually vexatious and often born of a lack of subject area knowledge. The article that lfromnj has linked provides plenty of examples of this (European data protectionism policy is particularly bizarre and arbitrary), and even Nathan does not deny that this is the case, even though for reasons that I don't really understand he seems to think that it is a good thing.

It's less that I think the EU's policy approach is good and more that I think the American policy approach of favoring and coddling this sector's economic and policy preferences well beyond what its social utility and contribution to human flourishing can justify is even worse.

AI development is hopelessly entangled with a deranged suicide cult. Half of this cult looks forward to the day in which, as they see it, we will all be "liberated" from the "tyranny" of physical existence and all be severed from our physical bodies, to be turned into bits in a machine where AI is constantly feeding our (simulated, digital) dopamine receptors with fentanyl-potency hedonium. The other half looks at the concerns about existential risk and says, Well Actually that's a good thing, and the sooner it exterminates us to turn the planet into computronium the better.

I'd agree with this more if people like Sam Altman and Eliezer Yudkowsky weren't a bunch of sad LARPy loons who, so far, are good mostly at generating ad copy and helping college students evade plagiarism raps. This might change in the future, but for now I think the dangers of AI are a new face of the stupidest and most craven parts of American capitalism generally, rather than something categorically novel.

[CumbrianLefty]

Absolutely deranged technoutopian framing. Europe hasn't so much "given up on tech" in that "tech" is an obviously good thing that it just doesn't want to do as it has "given up on tech" in that its governments and bureaucracies have decided that "tech" is something whose proper role in an economy and a society should be soberly and prudently assessed just like everything else. If European policymakers are triggering people who think that "tech" is some sort of killer app for human suffering as badly as they apparently are, then they must be doing something right, even though a lot of these regulations do seem badly written or badly implemented.

That is one way of putting it, yes.

(and the UK isn't much better)

[dead0man]

I've been hoping for years that some of these tech companies would do as Xahar suggests and just pull out of Europe.  Eventually the European voters will fix their bureaucratic nightmare, or not....it doesn't really matter to the rest of us (other than your stupid rules making it a slightly worse existence for everyone else).

[Vosem]

At least for social media, the US Supreme Court is hearing multiple different free-speech cases next term (Murthy v. Missouri on government jawboning, and NetChoice v. Paxton on voluntary moderation policies on social media) where the red-state position is essentially fundamentally incompatible with European law; a victory for the "red" states in the NetChoice case is not considered very likely but it's hard to see how it wouldn't immediately precipitate a schism between the European and US Internets if it comes to pass (the Fifth Circuit's view, if adopted, would mean that social media platforms active in the US would under normal circumstances not be permitted to follow European free speech law under the Court's interpretation of American constitutional guarantees*), and even Murthy -- where a red-state win seems likelier than not although not guaranteed -- would severely hamper any American governmental efforts to cooperate with European free-speech laws.

*I don't think the red states are likely to win this one at SCOTUS, although it does not seem impossible, but it would be fascinating to see what happens if some state Constitution is interpreted to protect free speech to this degree, which I think is a very plausible outcome in any number of red states, particularly Texas.

AI development is hopelessly entangled with a deranged suicide cult. Half of this cult looks forward to the day in which, as they see it, we will all be "liberated" from the "tyranny" of physical existence and all be severed from our physical bodies, to be turned into bits in a machine where AI is constantly feeding our (simulated, digital) dopamine receptors with fentanyl-potency hedonium. The other half looks at the concerns about existential risk and says, Well Actually that's a good thing, and the sooner it exterminates us to turn the planet into computronium the better.

I'd agree with this more if people like Sam Altman and Eliezer Yudkowsky weren't a bunch of sad LARPy loons who, so far, are good mostly at generating ad copy and helping college students evade plagiarism raps. This might change in the future, but for now I think the dangers of AI are a new face of the stupidest and most craven parts of American capitalism generally, rather than something categorically novel.

This view feels born out of a belief that things will stay the same forever; in fact AI is progressing pretty rapidly (over the past month the change in the quality of images which can be generated by an ordinary person has improved pretty dramatically, for instance). At the same time, we're running out of high quality data for AIs to use in the relatively near future, although more recent models have been able to get better performance out of less data, and it's already the case that OpenAI is using roughly 20% of the world's GPUs; to use many more it would need to embark on a program of building physical infrastructure, and creating factories that manufacture computer chips is really difficult (Morris Chang is really nowhere nearly as famous as he should be).

This post was originally going to heavily cite the article The Transistor Cliff in Asterisk Magazine arguing that many particular kinds of material science progress that are fueling improved performance in computing are coming to an end, and which had the best explanation I've read of arguments about whether or not Moore's Law still applies, but the website is currently down, so I'm going to leave this link here and encourage anybody who comes across this thread who is curious about the subject to read the piece if they can access it.

[Vosem]

Had to futz around with archive.org because the website is having issues, but I really strongly recommend The Transistor Cliff, which is by far the best piece I've read meant for a general audience about the technical barriers to general AI that exist today. While Moore's Law continues to hold for the moment the way it is usually phrased, of transistors per chip, several important correlates no longer hold (since about the mid-2000s, it has not held holding price constant -- chips with more transistors have gotten more expensive -- nor has it held holding power constant -- with the failure of Dennard scaling chips with more transistors have not required less power to use.*

Anyway, the thesis of the article is that advances in memory bandwidth are going much slower than the advances in how many computations can be performed by individual computer chips. Memory is stored separately from the chip actually doing the computation, and in the quite near future we're going to run into a problem where the memory required to train better AI models is just not physically available without some large technological breakthrough.

In other words, training GPT-3 took about 200,000 times as much compute as GPT-2, and GPT-4 probably took between 60-150 times more than GPT-3. In practical terms, GPT-2 could produce coherent sentences, but its output tended to degenerate into repetitive noise after about a paragraph. The much larger GPT-3 can reliably generate on-topic, sensible completions. GPT-4’s performance — on everything from programming problems to the bar exam — is even more impressive.

Looking at a longer time horizon, Epoch AI estimates that the compute used for training the state-of-the-art machine learning models has increased by about eight orders of magnitude (that is, 100 million times over) between 2012 and 2023.

...

Let’s say we want to no more than double that time — ever. And let’s say that DRAM bandwidth continues to grow at its current rate. In that case, it doesn’t matter how much compute we have. In order for the training run to be able to use all of its available compute without memory call times ballooning, compute cannot grow by more than 3 times by 2030 and 17 times by 2040. This is a much more conservative bound on AI compute growth than either what Moore’s Law for flops suggests (128 times by 2030) or what recent AI compute trends suggests (631 times by 2030). In a world where memory bandwidth (and time) is the limiting factor, we don’t even get one more order of magnitude of scaling growth in AI compute this decade.

This is not to say that we can't build an artificial general intelligence, or even that the current methods of scaling LLMs cannot possibly lead to an artificial general intelligence. (Indeed, if you do the thing I love doing on this forum and just extrapolate the trend-line out, then you may well reach the conclusion that a transformative AGI will appear in the 2020s. I think this would require a breakthrough in either memory storage -- the physical problem involved here is heat dissipation and a breakthrough in materials here might result in many new technologies -- or finding a technique for training AIs using much less compute than is used currently -- but "eh, the straight-line trend will just continue" often works much better to reason about the future than people like.** More subtle approaches that take into account slower rates of progress in memory bandwidth still usually forecast artificial general intelligence achieved within the 21st century, with around a 50/50 chance of it happening by 2040; this is also roughly what is forecasted by prediction markets, for whatever that's worth.

Still:

Moreover, the limiting factor in stacking more and more layers of memory, or packing circuit elements denser, is heat. Memory requires power to store information, even when it’s not “on,” and it dissipates heat. Today, even going beyond 12 layers may be infeasible due to heat constraints. Memory is especially sensitive to heat, because at higher temperatures, thermal noise can degrade stored data. Faster degradation means the data needs to be refreshed more frequently — but refreshing also generates waste heat! So there’s a vicious cycle where overheating leads to even more overheating.

Improving heat dissipation is an active area of research, and so more heat-efficient memory designs may be invented in future years, but scaling up memory bandwidth above its current slow trajectory is likely to continue to be challenging.

Sounds very hard to me.

*The old pattern, Koomey's law, was that the amount of power required to perform a computation halved about every 18 months. This was possible because of Dennard scaling, where the amount of power a transistor requires was correlated with its volume (power/volume = power density; power density remained constant); Dennard scaling failed circa 2005.
**Hell, my political philosophy -- of Hayekism -- has as its central point that the economy will grow given particular conditions but that predicting how it will do so, in a way useful for a central planner, is not even possible. "There will be a technological advance in this space, because the straight-line trend predicts it" is a correct take really frequently.

[oldtimer]

Productivity in Western Europe and Japan in lower today than it was in 2000.

And North America has also seen very low productivity growth too.

Basically technology and products that where invented since the end of the Cold War are actually less productive than the one they replaced, we've gone on the wrong path on the "tech tree".

[Okay, maybe Mike Johnson is a competent parliamentarian.]

I'm going to bow out of this conversation because I'm realizing I can't quite articulate the points I'm trying to make in a way that doesn't sound crankish, but I'll probably resume discussing this subject on the secular blog at some point.

[ingemann]

I've been hoping for years that some of these tech companies would do as Xahar suggests and just pull out of Europe.  Eventually the European voters will fix their bureaucratic nightmare, or not....it doesn't really matter to the rest of us (other than your stupid rules making it a slightly worse existence for everyone else).

Maybe you should ask yourself why they don’t boycott Europe, and try think about it instead of coming with a tired cliche. Play the devil’s advocate to your own views.

Beside that I hate these kind of article because they don’t serve to give us a real idea of whether “EU has given up on tech”. Usually it just mention that the big well known consumer brands are mostly American. But it doesn’t tell us anything about how the European tech industry look and how tech is used in the average European’s life compared to an American’s daily life. I must admit I can’t answer that question because like most people I don’t have a in-depth knowledge about the daily use of technology elsewhere*, and honestly unless people have lived elsewhere for years, I doubt they usually lack that knowledge too.

* As example if you visit Germany, the West Germans hates credit cards seeing it as a way for the state to observe them, while the East Germans don’t gives a sh**t and widely use them. That was something you would rarely find out unless you visited Germany.

[dead0man]

I've been hoping for years that some of these tech companies would do as Xahar suggests and just pull out of Europe.  Eventually the European voters will fix their bureaucratic nightmare, or not....it doesn't really matter to the rest of us (other than your stupid rules making it a slightly worse existence for everyone else).

Maybe you should ask yourself why they don’t boycott Europe, and try think about it instead of coming with a tired cliche. Play the devil’s advocate to your own views.

because they make more money off of you than it costs them*, for now.  Why else would any business be anywhere?  What other reason would there be?



*or they think being in the market now will bring large profits in the future

[ingemann]

I've been hoping for years that some of these tech companies would do as Xahar suggests and just pull out of Europe.  Eventually the European voters will fix their bureaucratic nightmare, or not....it doesn't really matter to the rest of us (other than your stupid rules making it a slightly worse existence for everyone else).

Maybe you should ask yourself why they don’t boycott Europe, and try think about it instead of coming with a tired cliche. Play the devil’s advocate to your own views.

because they make more money off of you than it costs them*, for now.  Why else would any business be anywhere?  What other reason would there be?



*or they think being in the market now will bring large profits in the future

There’s also a third reason, the main model of most of these companies is to already dominate the existing market, which make competition unlikely. It’s hard for a new Facebook to arise because everyone who want to use Facebook is already on Facebook, and Facebook only make sense if there’s a lot of people on it. But if Facebook leave a market someone else will fill the niche and make it next to impossible for Facebook to reenter, but it will also create a competition which can compete with in other markets. Most of the big American consumer IT brands are not really that hard to replace if they left a market.

Of course a difference why European companies didn’t rise up to compete in social media niche and the main problem in the first place is pure economics. The biggest European software company is SAP and it’s the fifth biggest in the world, most of you have never heard about it because it’s mainly geared toward commercial and industrial customer, meaning paying customers. American companies on the other hand ran or run with deficits for decades, focusing on growth and financing themselves through investor money, a few European companies did something similar but the end goal there was to be bought up by a bigger actor, European software companies focus on surplus and as such focus on paying customers like commercial actors, industrial customers, and states. Chinese companies have functioned in similar manners to the American ones, and this have allowed both them and USA to dominate the consumer niche.

[Zinneke]

I think the EU could solve a lot of problems if they regulated so that e-ink would overtake regular screens by 2030. E-ink would solve most of what's wrong with the internet's pro-ADD pro-hatred pro-dopamine seeking culture. In fact the worse thing to happen to this place is receiving notifications when quoted.

[CumbrianLefty]

Just looked up e-ink, an interesting hypothesis.

Is that why this site's (main) background colour is what it is?

[Antonio the Sixth]

lmao, I live in Europe and I enjoy tech just fine. I also enjoy having some ability to ensure my data isn't constantly harvested by malicious actors, and soon hopefully the fact that my social media aren't flooded with disinformation. It's absolutely hilarious that you have a bunch of Americans complain about the continued ensh*ttification of their online space, but when someone else tried to do something about it, it must mean they hate tech and want to destroy it.

[Zinneke]

Just looked up e-ink, an interesting hypothesis.

Is that why this site's (main) background colour is what it is?

No but it is the only site that works well with e-ink (and that's a good thing).

[Zinneke]

lmao, I live in Europe and I enjoy tech just fine. I also enjoy having some ability to ensure my data isn't constantly harvested by malicious actors, and soon hopefully the fact that my social media aren't flooded with disinformation. It's absolutely hilarious that you have a bunch of Americans complain about the continued ensh*ttification of their online space, but when someone else tried to do something about it, it must mean they hate tech and want to destroy it.

I agree with you Antonio

However there’s a difference between resisting the corporate gangsterism of neo-liberal USA and its hard on for Big Tech destroying the internet

And endorsing the way legislation is written in the EU, a model of ordo-liberalism designed to favour the people (read : multinationals) who can afford a “GDPR expert” rather than having a range of actors actually enforcing regulation

EU legislation does stifle innovation if it is too straightjacketed. The noises coming out of some EU legislators mouths on AI is really troubling : I really don’t care if a range of bullsh**t jobs become obsolete, I care about retraining the people who lost those jobs in doing something fulfilling.

It’s just sad one cannot have a debate about this without resorting to the awful American way of doing things as a counter-example. And even then the “California effect” is just as salient as the “Brussels effect”, its just that the EU legislators are so far up their own arseholes they think they rule the world by creating complicated legislation designed to create a ruling class of lawyer-types to obfuscate any strong political issue into a debate about the margins of some technical regulation.

[Blue3]

Productivity in Western Europe and Japan in lower today than it was in 2000.

And North America has also seen very low productivity growth too.

Basically technology and products that where invented since the end of the Cold War are actually less productive than the one they replaced, we've gone on the wrong path on the "tech tree".

Isn’t the average worker something like hundreds of times more productive than someone in 1920?

Which led to burnout, and poor mental health, among other things.



Anyways, while the US is leading the world in tech development and innovation, the EU seems to be leading the way in terms of regulation.

[ingemann]

lmao, I live in Europe and I enjoy tech just fine. I also enjoy having some ability to ensure my data isn't constantly harvested by malicious actors, and soon hopefully the fact that my social media aren't flooded with disinformation. It's absolutely hilarious that you have a bunch of Americans complain about the continued ensh*ttification of their online space, but when someone else tried to do something about it, it must mean they hate tech and want to destroy it.

In all fairness we don’t know what technology the Americans use, which is so commonplace that they don’t find it worth mention, maybe they have personal teleporters and robot servants. Through I must admit when I hear American tell horror stories about DMV or IRS I must admit I suspect that Americans make far less use of their super technology than many other countries does, because I do most of my interaction with civil services over the internet, I log in 13 marts 2024 to see what taxes I have paid in 2023, make notes if I have earned more or have deduction and either get money back or get told I owe them more money and get a offer for a installment arrangement. If I lose my physical drivers license, I order a new one and get a email (on a specific separate secure email only for public information, but which also include information from some private actor like my bank, private pension, union, and employer*) with a temporary one, which I print out or keep on my phone. I think the last time I had to meet up in person was 3 years ago, when I had to get a new passport. I walked down to the municipality office, took a number waited half a hour, proved I was who I said I was, they took a picture of me and I got my passport in snailmail later. If I was unemployed and got benefits, I would also have to meet up once in a while to ensure I was still in the country.

So in general compared to what I hear from Americans, their civil services sounds like they’re still using fax machines, and make little use of what new technology offer in saved time and improved services.

* when it was introduced it pretty much killed the postal service, whose number of employers have been reduced to 25% of its staff 15 years ago, and mainly is a packet delivering service today.

[oldtimer]

Productivity in Western Europe and Japan in lower today than it was in 2000.

And North America has also seen very low productivity growth too.

Basically technology and products that where invented since the end of the Cold War are actually less productive than the one they replaced, we've gone on the wrong path on the "tech tree".

Isn’t the average worker something like hundreds of times more productive than someone in 1920?

Which led to burnout, and poor mental health, among other things.



Anyways, while the US is leading the world in tech development and innovation, the EU seems to be leading the way in terms of regulation.

I think there is a growing feeling that the goldilocks period of tech was the late 1990's - early 2000's and that something went wrong since.

[lfromnj]

Why should Europe want tech companies ?
Because they have created some of the highest paying jobs in America and int he entire world for millions. Meanwhile in Europe the same job would only pay a third of what Americans get paid.