[pembrook]

My feelings about this rest of the scope of liability. From my understanding, Google is now liable if making false claims about a personal/business reputation. I like this idea in theory (key word).

However, I can easily see the slippery slope where in practice this means providing any AI response becomes too risky, and it becomes another money club used to extract wealth from big tech due to the current hysterical anti-AI moral panic.

Which would ultimately kill the ability of the German people to get access to competitive AI models.

I'm sure many anti-tech/anti-civilization doomers on HN will cheer this on. However, in reality it would do nothing to stop Germany falling behind, and continue its economic malaise/low productivity growth and social welfare collapse. When things in Germany get bad, Germans historically have tended to...ummm...cause issues for Europe. I would not take this lightly given the current rise of more polarized political rhetoric and German economy pivoting hard into weapons manufacturing.

[sham1]

The slippery slope of being held liable for an AI overview leading to neo-Hitler is certainly a take, and a ridiculous one at that.

It's one thing for an LLM to get things wrong. This case is not that though. And if big tech can't make sure that their models don't libel companies and individuals then good riddance. Whatever diminutive economic advantage one can get from "competitive LLMs" probably isn't worth it anyway, especially with all of the other disadvantages of these "competitive LLMs" compounding on our societies and the planet, as the resource usage of the data centres necessary to run and train them exacerbate the ongoing climate crisis.

[pembrook]

You've formed a strawman out of my argument.

My point is Germany further rejecting participation in the the next wave of technology (as they have done for 50 years) is going to NOT improve the trajectory they are currently on...which is bad.

The neo-Hitler and re-militarization thing is already happening. AfD leads in the polls, they're opening talking about remigration of immigrants, Germany now produces more ammunition than any country on earth. AI has nothing to do with it.

And, given I'm speaking to one of the the anti-civilization/anti-tech doomers of HN as predicted, do you think Germany rejecting technology again is going to improve the "climate crisis" or hurt it? How is the de-commissioning of Nuclear working out there?

[kevinxsun]

It is not anti-AI, it is anti misusing AI. Clearly you are trying to manipulate the situation here. If your AI is half baked and produce garbages, you should be liable for the damage of those garbages, that applies to any product or service on this planet, not only just AI.If your product or service is defect, that causes harm, you are liable.

[pembrook]

I agree with this, in theory.

But to loudly proclaim your love for this idea as most have here seems a bit premature given there are potentially massive downsides to applying legal liability to an AI model, if not done extremely carefully.

I'm just trying to move a bit beyond the emotional first order thinking. You might see this as manipulation if you love just swimming in emotional vibes, but I find this website becoming quite boring lately due to the lack of debate beyond this first-order thinking.

As someone living in Europe, watching a bunch of rich American engineers fanboy European over-regulation which they aren't actually being subjected to, is super cringe. It's the epitome of the principal-agent problem.