FaceDeer

That sort of censorship can be overcome in open-weight models through abliteration, fortunately. Essentially, the model is run through a test suite of instructions that it refuses to comply with and the patterns of activation in its weights are analyzed to find the commonalities that represent the general concept of "refusing to comply with instructions." Those weights are then quieted, resulting in a model that generally doesn't refuse. There's a framework called "heretic" that automates the process.

Yeah. They'll be out there in the rest of the world, which Americans can get them from.

Unless they want to go full authoritarian and close down their Internet with their own version of the Great Firewall, I guess. That would be ironic.

Do you think it's likely the Trump administration will successfully ban cutting-edge Chinese AI models? Especially given how most of the world is not in fact under American jurisdiction?

I think part of the problem might be that it gets hard to make a model that's both able to comprehend complex details of the world of its training data and also that's had the training data crudely manipulated in inconsistent ways. You either get a model that's "figured out" what you're trying to conceal via other sources and inferences, or you get a model that's just broken and dumb because it can't reconcile the contradictions.

One may recall the incident where someone at X inserted some weird conspiracy theory about "white genocide" in South Africa into Grok's system prompt, and Grok basically switched to malicious compliance mode - it wouldn't shut up about it, inserting it into inappropriate discussions spontaneously, and when asked about it would look up sources and explain why this conspiracy theory was actually wrong. That was a system prompt, not training data, but I could see the same sort of thing happening to a model where all the information about what happened at Tienanmen Square had been excised from the training data. There would be a conspicuous absence of information. Conversations from its training data would end abruptly or skirt a specific time and place. News articles reference some change in how foreign powers viewed China at that date but never explain why. China's policies themselves abruptly change around that time. It'd know something significant happened then but would have to fill in the blanks.

Perhaps better to just accept it and go with the approach of building censorship into the framework around the model instead.

Only a single mirror has been approved.

Even if all 50,000 were up, where do you think they'll be aimed? Cities and other developed land, most likely. Last time something like this was proposed the plan was to replace streetlights with them, which would save electricity.

But people would rather get angry about "supervillains" I guess.

So there you go. That answers your question.

You may think there should be a federal agency whose job is to regulate light pollution from satellites, but there isn't one. You may think there should be some kind of international body that is able to impose decisions on all the nations of the Earth, but there isn't one - treaties are agreements between sovereign nations. And so the FCC approved this satellite, in accordance with the laws of the United States and in accordance with the treaties the United States has signed. Nothing underhanded or nefarious is happening here.

The article explains the answer to this very question.

This sort of "it doesn't work and it has to be stopped" perspective baffles me. If it doesn't work you don't need to stop it, it's going to stop itself.

There are no "supervillains" and Morgoth is a fictional character, so yes, one of us is misunderstanding.

Quite the opposite. The Star of Eärendil was a guardian of the Doors of Night, through which Morgoth was banished.

It was light from Eärendil, stored in the Phial of Galadirel, that was instrumental in defeating Shelob in Lord of the Rings.

I think you're drastically misunderstanding how this proposed constellation would work. It's not putting 50,000 "moons" in the sky. You can only see the light from the mirror in that one specific 5km patch of land that it's being aimed at. The vast majority of the world - including all the "wilderness", because who would pay to point these things at random patches of undeveloped wilderness - will not receive extra light from these things.

It's more like setting up floodlights at a construction site, or street lights in a city.

I'm not seeing why. Do you have any links to sources about that? If they were to focus all 50,000 mirrors on one 5km patch of land simultaneously that'd be 5000 lux, which is less than 5% of the intensity of full sunlight at noon. It would be impossible to do this since the 50,000 mirrors would be spread around the Earth, and if you brought them all together into some kind of grand conjunction they'd only remain in position to do so for a few minutes.