Trump has this utterly fundamental belief that you're either a "winner" or a "loser." The moment you give something they want, that makes them the winner and therefore you are the loser. He doesn't care about the numbers, he just cares about whether he's winning.

There actually was a live action Robocop TV series decades ago. Only ran one season but I remember it being rather good.

You may not have tried out modern abliterated models. GPT-OSS is almost a year old now, I assume the abliterated version you tried was pretty old too? The technology has improved significantly, it's much more focused on just targeting refusals. A well-abliterated model benchmarks basically the same as its original version these days.

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.

Breaking news: Screaming toddler repeats the same threats he's made numerous times before.

Frankly, this kind of meddling just makes it more appealing to use Chinese models. Chinese models are often open-weight, meaning that they're basically immune to foreign meddling - if China decides tomorrow to "cut you off" you just need to find some other provider that's offering those models and buy access from them instead. Or if you're a big enough organization, just rent or buy some hardware and run it yourself.

This sort of meddling shows that the US government is eager to weaponize access to AI, and if you're dependent on American models there are no other places you can run them.

Have to admit, I've completely forgotten about Pear Harbor.