Kaligalis

Ad blocking is too important to depend on a browser vendor for it.
uBlock Origin kept updating basically just-in-time in the great YouTube ad war. No browser vendor can or even just tries to match that level of dedication and quality. Ad blocking needs to be pluggable - also because of SponsorBlock which is essential for YouTube.

Yeah, but it does run uBlock Origin, renders web pages just fine, can do WebRTC, and even has support for bookmarks and pinned tabs. What else do you want from a web browser?

Chrome’s advantages were in the security and performance departments. But Firefox isn’t that much worse.

That’s the only reason, I use it. And I knew Firefox would be the only one remaining when I switched to it the very moment, Chromium stopped supporting uBlock Origin.

That sounds a bit like it conflicts with the actual job of the LLM. And the resulting watermark would be way too fuzzy to be actually useful for flagging anything as AI-generated.
Are they just trying to check a box on some compliance checklist?

When other corps see that Facebook ads are unblockable they will copy the mechanism and uBlock Origin will become useless.

uBlock Origin really needs to find a way to define content-based rules for sites which use DOM randomization. After all, the ads still have places where they can go and ad content is structurally always very different from actual content.
If they let Facebook win, YouTube will do it next. And a year later, uBlock Origin will have stopped working on half the web.

Not really news. France has these problems every summer since decades. It’s a common problem of river-cooled power plants. The obvious solution is to build them at the coast - like in Fukushima.

This has been solved for cars. They can kill people and depending on the cause, the manufacturer, the driver, the passenger, the victim, or even no one can be responsible. AI is just yet another technology with potential for hidden safety issues and other faults.

Everyone is like “damn AI is making kernel be too big!!111” while Linus is like “Tons of genuine fixes and improvements found/made with AI lead to a lot of commits being the norm. Lots of review work, but definitely worth it.”

Those hidden markers will be very hard to hide in code review in an IDE which highlights invisible Unicode code points. For code, adding watermarks is likely to cause bugs. For text, Chinese models are as good - so just switching to them solves the issue.
I think, Anthropic doesn’t want to perform economic seppuku.

Has to be fake. Russia is out of missiles since years.

“Nobody Asked for AI” is a pretty stupid argumentative approach.

• It’s obviously wrong because a lot of people ask for AI which actually works. And I myself use AI daily because it does work for some things better (and yes, sometimes that just means faster and/or cheaper) than the alternatives right now.
• If actually nobody had asked for AI, that would still be a null-argument because AI is pretty damn useful regardless. There was a massive resistance against locomotives, cars, power looms… technology in general. Some new technology was outright wrong. But after all, people do live longer and less children die per capita now than before the industrialization. Not knowing that it will be useful doesn’t imply it isn’t useful. You are using technology which simply didn’t exist a hundred years ago right now. Technological advancement is actually a proven-good thing.
• Arguing against AI in general completely misses the point. What people actually hate is US AI companies. And rightfully so. The self-licking ice cone of circular funding brought us the RAM shortage, massive power grid destabilization, local environmental problems because of reckless construction and operation of data centers (on paper, data centers are the absolute cleanest thing you can build; but they use dirty gas turbines because the grid can’t sustain them; and somehow they also manage to waste massive amounts of water - no idea, how though, as all professional cooling tech uses closed loops).

That Kurzgesagt channel doesn’t sound like a natural human voice to me either. YouTube might be right.

To be fair, oligarchs and three-letter-agencies are pretty much above the law everywhere.

You literally can’t control the proliferation of open models made outside your jurisdiction.

Just a hundred years ago, we had one floating point operation per second in mechanical calculators. We got roughly 100 TFlops (fp16; probably much more accurate than human brain neurons) in a high-end gaming GPU now. That’s 14 orders of magnitudes in a hundred years. I guess, 4 more orders of magnitude in another hundred years should be possible. We don’t even have fusion power yet. A lot changes when power becomes a practically unlimited resource.

China becoming an industrialized country and the people there slowly getting to a higher standard of living is actually a good thing for us humanists for whom all humans have equal rights.

The problem is just that we need to migrate our thinking from a colonial exploitive mindset to a more sustainable one. The US can in principle just rebuild all the industry it lost and everything is fine until the oil finally runs out.
But the EU needs to start accepting that they are not rich on natural resources and that the countries which are, are becoming less reliant on imports. That means that the EU will run out of things to give in exchange for resources. If resources can’t be bought anymore, what is already there needs to be recycled to almost 100%. And no bloc can long-term be dependent on energy imports because when everyone is industrialized, everyone has basically unlimited use for energy.

The solution is self-sufficiency on a bloc level. China, Russia, and the US should have no problems being their own blocs. Everyone else needs to overcome the disputes with their neighbors and create or join a bloc.

To be fair: Yes, we do have that pipeline. But we do not have long-term strategic government funding of strategically important industries - especially not Dutch ones. China tries to plan a century ahead (and seems to actually be able to plan a few decades ahead). No western government seems to plan further ahead than an election cycle.

A centralized guaranteed stable government has different opportunities to plan ahead than a government which has to primarily care about being reelected. Our simulation of democracy isn’t free. The massive election theater and lobbying comes with huge price tags (which are mostly invisible to the public until no one can afford to have kids anymore).

The Dutch government forced China to make these machines themselves by enforcing the embargo demanded by the US.
Now, China does exactly that. Then they scale up until they have strategic overproduction. Then the surplus gets thrown on the global market for cheap. Then ASML dies.
Goal: Prevent China from getting the tech.
Actual effect: The west becomes dependent on China for the tech.

The article is about AI worms. And the thumb is what an AI image generator thought, AI worms could look like.