[Herz]

I stated quite clearly that not every regulation is bad. But it seems that you want to hear that every decision made by the EU is right. I'm sorry, but I'm not a religious person. And I think self-criticism is a great privilege of democratic (not dictatorial) countries, so let's use it.

> Ensuring that AI can't be used to make life or death decisions if its decision-making can't be explained (which the AI Act boils down to)? Ensuring there is competition on, for example, railway operations?

It's such a naive question that I can't understand how you can take it seriously.

Just because you can explain how you arrived at a specific decision does not mean that failure does not exist. Every machine is fallible. Every human is fallible. Moreover, you cannot determine decision-making made by humans. So how can you trust humans? Why should you trust them?

https://en.wikipedia.org/wiki/Suicide_by_pilot

I would like to see the data, not the social or individual biases. It's only a matter of "when" AI will prove to be safer than humans at performing task X. I find it absurd to deprive ourselves of such an advantage, supported by data, just because our understanding isn't absolute.

Can we prove the safety or determinism of what we use or do on a daily basis? I doubt. Shouldn't we experiment with physics because our understanding is limited, and we might accidentally create a black hole? I doubt.

Also, I find it such a generic definition... Google Maps implements AI, and accidentally sends you into a ditch. What do you do? Ban AI from Google Maps? What doesn't put people's lives at risk?

I totally understand the skepticism and fear. The risks, etc. But I'll leave it to the fortune tellers to pass judgment before it's even "a thing".

> It's such a common refrain that EU is just stifling competition with "regulating everything", but quite oftne EU regulations are actually forcing competition where none was possible before.

Is killing the car market "forcing the competition"? How?

[sofixa]

> I stated quite clearly that not every regulation is bad. But it seems that you want to hear that every decision made by the EU is right. I'm sorry, but I'm not a religious person. And I think self-criticism is a great privilege of democratic (not dictatorial) countries, so let's use it.

But you still said that you think most of the EU's are bad, so I'm opening the discussion with multiple that I consider to be good.

> Just because you can explain how you arrived at a specific decision does not mean that failure does not exist. Every machine is fallible. Every human is fallible. Moreover, you cannot determine decision-making made by humans. So how can you trust humans? Why should you trust them?

Of course not, but being able to explain the decision, and thus prove that it is wrong, and have humans being able to correct it, is good. It means that stuff like United Healthcare Group using algorithms to decide if care can be paid for, with a terrible failure rate, and employees just shrugging "computer said no" cannot happen in the EU. The fact that this kind of things are considered as "EU is killing AI with too much regulation" is really concerning to me.

> Is killing the car market "forcing the competition"? How?

How is the EU killing the car market, exactly?

[Herz]

> But you still said that you think most of the EU's are bad, so I'm opening the discussion with multiple that I consider to be good.

I understand your point, but I see no reason to invest time defending the EU's positive aspects. What's the point?

> Of course not, but being able to explain the decision, and thus prove that it is wrong, and have humans being able to correct it, is good. It means that stuff like United Healthcare Group using algorithms to decide if care can be paid for, with a terrible failure rate, and employees just shrugging "computer said no" cannot happen in the EU. The fact that this kind of things are considered as "EU is killing AI with too much regulation" is really concerning to me.

I don't see why "asking for less regulation" concerns you. The EU seems to listen to people like you, not people like me. I should be the one who's concerned, haha. I'm worried because bureaucracy is a slow-acting cancer. It's a process that's easy to start but incredibly difficult to stop or reverse.

The problem with bureaucracy, regulation, and welfare is that they all come with a price. Increasing costs require a strong, cutting-edge economy to sustain them. Yet, no one seems to be concerned. In the US and China, new technologies are constantly being created, while in Europe, innovation is stagnating. No one seems to care that Europe's wealth is fragile, based mainly on "old" companies or banks.

Of course, no one is against welfare; my concern is its unsustainability. As an Italian (living elsewhere in Europe), I find the situation worrying. The demographic decline is dramatic, and pension and healthcare costs are skyrocketing. In Italy, a worker under 40 often earns less than a retiree. With such a sharp demographic decline, retirees have enormous political power.

Europe is aging, and so is its appetite for innovation and risk. Yet, we keep adding costs upon costs. Even if the goals of initiatives like GDPR, the AI Act, and the Green Deal are "right", we can't deny that they come with a price. This added cost inevitably makes companies less efficient in Europe. This is a simple consequence. Can we truly afford this?

How long can we keep going? The rope will break sooner or later. And why doesn't anyone seem to care?

> How is the EU killing the car market, exactly?

1) https://commission.europa.eu/topics/eu-competitiveness/dragh... 2) "The Draghi report: In-depth analysis and recommendations (Part B)" 3) Page 146

I think this report its quite comprhensive to state what its not going that well in EU.

I dont agree with everything in the document, but i think its a good starting point.

[sofixa]

> I don't see why "asking for less regulation" concerns you.

Because the "less regulation" is in response to the EU saying you can't have algorithms making life or death decisions if they can't be explained and can't be escalated to a human. People are literally asking for companies to be able to shrug behind "computer says no" with no recourse. We have the UK Post Office scandal for a closer to home example on why this is a terrible idea. "Less regulation" here would be plainly terrible for everyone.

> No one seems to care that Europe's wealth is fragile, based mainly on "old" companies or banks.

Along with migration, it's probably the two most discussed topics. Funnily for it too, everyone says "nobody cares", yet it's literally among the most discussed things.

> Even if the goals of initiatives like GDPR, the AI Act, and the Green Deal are "right", we can't deny that they come with a price. This added cost inevitably makes companies less efficient in Europe. This is a simple consequence. Can we truly afford this?

I get what you're saying, and there's a point at which I would agree; but I also fully consider that allowing companies to let people die and hide behind "The Algorithm" is something so fundamentally wrong, that we cannot (humanely) afford not to have regulations against it.

> In the US and China, new technologies are constantly being created, while in Europe, innovation is stagnating

Because you're comparing massive economies with lots of capital to burn, vs a loose collection of much smaller countries. There is tons of innovation in various European countries, it's just of different types, and doesn't scale nearly to the same extent. And that is a problem (because, as you said, a lot of the economy is reliant on big old players, which isn't necessarily bad, but is lacking in economic diversification).

> As an Italian (living elsewhere in Europe), I find the situation worrying. The demographic decline is dramatic, and pension and healthcare costs are skyrocketing. In Italy, a worker under 40 often earns less than a retiree. With such a sharp demographic decline, retirees have enormous political power

It's the same in France too, and it is indeed worrying. Public budgets are getting increasingly more complicated to balance.

But, allowing companies to deploy AI to make life or death decisions won't change anything around this. Allowing them to harvest personal data without even knowing what they have won't change anything around this either. Allowing gatekeepers to stifle any possible competition (not having DMA/DSA), same thing.

The biggest changes needed are capital investments to help the tons of startups all over Europe scale; and complex policies to help minimise the demographic collapse. Some of it is natural and nothing can be done about it (if a couple doesn't want kids, no amount of aid is going to change their mind), but for others it's a matter of being unable to afford (more) kids.

[Herz]

> Along with migration, it's probably the two most discussed topics. Funnily for it too, everyone says "nobody cares", yet it's literally among the most discussed things.

Its disscussed here, still nobody is acting. This is a bubble.

> I get what you're saying, and there's a point at which I would agree; but I also fully consider that allowing companies to let people die and hide behind "The Algorithm" is something so fundamentally wrong, that we cannot (humanely) afford not to have regulations against it.

This sentence is fundamentally wrong, no one is dying. And for me, it perfectly sums up the issues we're discussing.

We've reached the point where if there's a risk of something happening, no matter the probability neither the magnitude, something must be done. Even if the solution is totally destructive, inappropriate for the problem, etc. Or even worse, deciding when the problem does not yet exist. Or the technology is still in its early stages. Like AI. This is what you are proposing. This is what I criticize.

Slowing down or stopping everything because MAYBE it's the right thing to do, MAYBE something we don't like might happen. This comes at a cost, especially if you apply this principle to everything around you in small doses. It's poison for productivity and efficiency.

I don't know if you are for or against nuclear power. I am quite pro nuclear power. But everyone knows about the European Pressurized Reactor (EPR) project, it is a failure in terms of costs and bureaucracy. China and South Korea are able to build reactors quickly and at low cost. The same EPR reactors built in China have low costs and short construction times (I am referring to the Taishan Nuclear Power Plant). The problem is exclusively European.

In the name of some ideology, we are destroying our productivity and efficency. Again. Why?

And I know very well that the answer is always the same. Safety. But it's just an excuse to sell you the services of yet another bureaucrat. There are very precise risk analyses that show nuclear reactors to be orders of magnitude safer than all other energy sources. So why this ideological obsession? Safety has nothing to do with it.

No one cares about risk analyses. Because the answer will always be “it's never enough.” But at what cost? Again, no one cares.

And thanks to this choices, in the name of safety, building reactors in Europe is difficult and expensive. But in the meantime, it is perfectly legitimate to build gas or coal-fired power plants.

Europe is full of this kind of hypocrisy.

[sofixa]

> Its disscussed here, still nobody is acting

No, it's discussed everywhere, at the EU and the local level. There has been plenty of action at various levels (like in France, under Macron first as minister of the economy and later president; and he's been decried and criticised a lot, but has also gotten a ton of reforms through).

> This sentence is fundamentally wrong, no one is dying. And for me, it perfectly sums up the issues we're discussing.

That's the point though. Literally the main thing the law does is that if the AI can make decision that can result in deaths, there should be a human escalation and its decision making should be explainable. That's it. If that's too much burden, something is wrong.

> Or even worse, deciding when the problem does not yet exist. Or the technology is still in its early stages. Like AI

But the problem already exists, again, cf. United Healthcare Group in the US. We know they're killing people and hiding behind a well known faulty "AI". We don't want that shit in the EU.

> I don't know if you are for or against nuclear power. I am quite pro nuclear power. But everyone knows about the European Pressurized Reactor (EPR) project, it is a failure in terms of costs and bureaucracy

If you're pro nuclear, you should know what the real problems with EPR are. The main are failures at EDF with the quality of their work, due to lack of qualified personnel, like welders. This has been well documented for Flamanville and Hinkley Point, and EDF has even written extensively about all the lessons learned from those disasters that have been incorporated. They even flat out say that Flamanville has allowed them to build industrial capacity and human know how to be able to build the next ones.

Do you have anything to back your claim that somehow bureaucracy is to blame? EDF are a state owned company, but I'm pretty sure that the British wouldn't stop yapping around if EDF were bungling Hinkley Point because of French/EU bureaucracy. There should be at least as much material on it as there are about the quality control issues, right?

> The same EPR reactors built in China have low costs and short construction times (I am referring to the Taishan Nuclear Power Plant). The problem is exclusively European.

Yes, because we stopped building reactors for decades, and nobody is around that knows the intricacies of that. Hence the investment in EPR, to improve on the failures at Flamanville, Hinkley Point, Olkiluoto, and be able to reliably deliver EPR reactors with predictable costs.