CMay

How about ask AI for all the times all the US presidents did something that went against the constitution. I think you'll be shocked. The court system being a bit slow is part of the point, because we're in an environment where China wants to use the speed of democracy as a weakness. Since you're on HN maybe you understand the concept of slow journalism or the benefits of anything being slow. Most people agree with that, having things be a bit slower is a useful check against errors. We understand that, the US was literally founded on that. China does not care. In that environment, we have to make compromises.

Being less predictable makes it harder for our adversaries to calculate. We're still good people, but we're in bad times. You don't need to understand. Through most of history only a tiny number of people anywhere in the world truly understood what was going on and what was necessary to respond to the times. This is true in science as well. How many times in history was it only a few scientists that had a clearer picture? Everyone thinks their religion is _the_ religion. The same can apply to news and opinions.

If you think the US is about Trump's ego, you still don't understand.

Yes and no. The US presidency is very different than the way it's reported on, but it is true that the executive branch is being more assertive to make up for the disfunction of the legislation. We're doing that intentionally and for good reason, to demonstrate to China that the US can act with speed as one small part of the overall deterrent against World War 3. The US is almost always hugely misunderstood, because there isn't any other country in the world that actually works exactly like the US does.

This tracks with my experience with their models. Nice work. It would be interesting to see what they could do if they optimized even more for moderation like this without going so far as to hurt the model performance overall.

I'm largely referring to the fundamental components that can be combined to arrive at a decision. Is this a real life image or text, or is it some sort of art or fiction? Is this in a context of historical events for education, etc? Is this sarcasm? Is this a real threat? Is this nudity? Does this contain violence and gore? Is this suggestive? How old is this character? Is this content sufficiently related to the content it's replying to, or is it spam? Does this response deviate too much from the input request? Is this a business logo?

There was this book called The Inquisitive Mood and it very much reminds me of the chain interrogation process you can do with models now.

This isn't exactly a single use case, it's more like there are use cases for image editing software that use generative models, there are use cases for local LLMs to get a sense for abstract user intent, or to check if a model's action is in line with policy and so on.

As I said, the Liquid AI models are already very good at a lot of this, but considering how good the models are at it, I'm curious how much y'all train on tasks that are moderation-adjacent. If y'all don't have any model that's really specifically trained for these kinds of tasks, it would be interesting to see how close y'all could get if you tried.

This is an actually important problem in the AI space especially for local development.

There are other moderation tasks that are harder, like "is this copyrighted?" or "is this a McDonald's logo?" because of how much copyrighted material there is and how many businesses there are, but it might be possible to get a good feel for "is this likely a trademark?", "is this a business logo?", "does this text seem to describe a popular character?"

To some extent you can combine a model like this with a good database to optimize the filters and identification. Then you can provide a few reference image or text examples that are closer to what you're trying to filter out and the model can say, "yeah, this content is pretty close to your examples."

Google and OpenAI have already released some of their open source safety models and they were really quite bad, from my testing. Yet, Liquid AI hasn't really leaned into promoting their models for moderation purposes despite being excellent. I know there is a lot of political intersection in moderation, especially in the past many years, but I think a model that is smart enough and excels at letting people apply their own rules with fairly good accuracy is important right now.

Did y'all do any particular work that makes these models great for moderation? One of the issues between cloud models vs local is that often people are trying to push the limit of what their hardware can do and it doesn't leave too much headroom for moderation to make the models a little more constrained for kids or schools to use locally, or just for family in general who don't want accidental unhinged results.

In the past I've noticed that some LFM were way better at moderation than much larger models. Not necessarily prejudging what is acceptable, but allowing you to interrogate content with a series of questions to customize the moderation. Other models I tested were far worse.

Decision models are like the next optimization in that direction, but none of the benchmarks for the popular models ever show how good they are at tasks like content moderation. The smaller and more efficient we can get that workload, the more that it can simply be on by default in a lot of use cases without feeling like you're giving up a lot of performance, creative freedom or accuracy. You can't have a high number of false positives or people would rather take control back.

We need local content moderation to be cool, efficient, accurate and configurable so that people can get the higher quality results which unsafe models tend to produce, while filtering out the unsafe results fast enough that it can even be done in an agentic way and fed back into the model to steer its results.

This would be amazing for making audiobooks more expressive and like real storytelling. I kept telling Audible they needed to invest more into premium audio experiences, kind of the way you had the old radio plays that people used to tune into. The War of the Worlds! We are seeing some of that with full casts of actors doing readings more.

At the same time, imagine if every book was like a living thing you could interact with and buying the book partially bought you access to the model that would tell you the story. The only way you would interact with that particular model, is if you bought the book.

It reads the book to you and you could chime in or it would know when to pause if there's an interruption without you needing to manually pause. You say hey wait, why is this character so angry? Then it kind of naturally reacts and continues the story. Maybe you come back to it 3 years later and it remembers you liked a certain part of the book and injects "here comes the good part!" It hears you start snoring and begins lowering its voice and whispering, but remembers you asked it to wake you up loudly if you start snoring, so it bolts out HEYWAKEUP!

For a lot of people books are a very solitary thing, though of course there are social clubs and book readings. Almost makes me think of those magical books you see fantasy wizards having, where the books themselves are sentient. It would be crazy if they released a The Princess Bride book and had the real narrator's voice and you could interrupt him and he could harass you about all your interruptions.

It wasn't that long ago where we were talking about having games and movies blend so the movies become interactive experiences. We're really getting closer to some of these static mediums becoming more like, trained behavioral meshes that have a personality and you can kind of naturally spend time with it. I'm not sure I'd be big on paintings on the wall coming to life and talking to you like in Harry Potter, but there's some promise.

Game worlds are an obvious extension, not just through NPCs, but if anyone remembers that game called Bastion where there was a narrator explaining the events as they went on. You could imagine a model that could watch you play and speak naturally, being funny or narrating. Almost something like Mystery Science Theater 3000 or how Beavis and Butthead would react to music videos.

I don't think those should be pervasive or the new norm, but there probably is an opportunity there and a market for it. Some kind of social-like presence that doesn't get tired, doesn't get bored, always has the same schedule you do, etc.

Hell, that reminds me of Seaman on the Dreamcast, too. SEGA could make a new Seaman and it would actually be awesome. Like a blend between Nintendogs, a fish, and some AI. There could even be charming goofy antics it tries to pull where it knows it's stuck inside a game and tries ridiculous things to manipulate you into letting it out. Some kind of Pinky and the Brain charm.

Anybody on board with this future?

If you can't profit off of your product sufficiently, because not enough people are interested in the experience that streaming provides, then you can no longer justify even making those games and it all becomes irrelevant. So, in the end you release your game to local hardware anyway.

They can't simply copy your game and reship it, because that would be against the copyright laws in most places where these companies make their profits from. Making something sufficiently new would be technically legal, but it would also take massive computing resources even with LLMs. That's either local compute time or cloud spend.

I think the bigger risk is just brand new games being generated of all kinds rather than the more expensive and legally risky full decompilations. Mass generation could start saturating the market and soaking attention. The low end indie market is already pretty saturated and in the end games have to be appealing, review sufficiently well and find an audience. Plenty of good games have barely sold any copies.

So almost a bigger question will be, how will AI impact the marketing industry?

This is not necessarily the case, as you still need to deliver an experience that customers enjoy and will pay money for. This means consoles that let them play games in their hands with the highest picture quality and lowest latency. No disruptions from poor wireless or significant interference in environments like apartment complexes.

Fully homomorphic encrypted processing may also become a norm and if that is combined with some games going back to being console only, there could be significant enough protection for games that contain trade secrets. That said, the days may be numbered for trade secrets at all given that better models may make today's clever trade secrets look juvenile.

At the same time, a lot of these games were already getting decompiled and inspected at runtime. That has historically been one part of the incentive to release games on console only or console first, particularly first party titles.

In any market where it's relevant to their business interests, they can still rely on the copyright laws and can set traps for any companies using decompiled code as part of their own paid and sold product. That whole process will become cheaper and easier to do as well, plus new approaches to slowing down or hindering the results of decompilation will become normal for games in mathematically provable ways. We've already seen approaches over the years like Denuvo.

Always online games themselves have benefited from some of their code being in the cloud and that will obviously be under fire more than ever with people writing their own servers, but even that might have novel approaches that can survive advanced tampering.

I've tried it some, only locally. I rarely use APIs, so can't say what the official inference API results look like. It correctly followed some prompts that almost every model I've ever tried performed poorly at, but failed in strange ways on prompts that other models handle with ease.

At high resolutions, structural integrity on people can definitely fall apart like older models, where you get one person turning into a long spaghetti spanning across the frame. It might be that you need to do grounding by generating some kind of scene reference at its stable resolutions first then having it use that for the larger resolution like older models too.

They say that their prompt "upscaler" model or system prompt are essentially necessary for the model, but my results have been hit or miss with that claim. Some basic prompts are better, some upscaled prompts are better. I think that if you gain an intuition for how the model understands things, you might get better results by default by hand crafting prompts specifically for it.

Some of the stylized art I'm getting out of it at higher resolutions are really nice and I prefer them over other models.

All of these models seem to have their own character in terms of system requirements, performance, prompt adherence, artistic outputs and so on. None of the models I have tried are a complete homerun, so maybe in this case the model feels a little inconsistent in some ways, but it's a model I will probably keep around.

The kind of stuff that people are doing with decision models today is largely the way I've been using Liquid AI's LFM models, as rapid iteration interrogation tools for atomic decisions. They aren't good at reasoning, but you can reason about what decisions you would need in order to come to a reasonably high quality decision about a thing.

Even though they weren't themselves decision models exactly, they were so fast that you could use them in a similar way. They obliterated Qwen and other models for a lot of tasks in that use case.

I'm not sure I would agree with some of the claims made in this article though. You can definitely do a lot of similar tasks to regular LLMs with decision models, you just have to chain the results. You can even technically infuse a broader perspective into each token choice or force certain context to have priority in the decision making of the next token.

> Again, Hitler didn't blitzkrieg GB because communism.

Of course not, he never planned to attack the Brits. They declared war on Germany after Germany attacked Poland. To be fair, they told him that's what they would do if he attacked Poland.

One point of attacking Poland even though they were anti-communist and inviting Russia to join in, is to disguise the true aim from Russia which was to invade the Soviet Union and destroy the Communist International. And they certainly eventually did invade.

Japan would have joined in, but invading Russia from their side was very unfavorable.

> Marxism is fundamentally a workers movement. The CCP is a lot of things, but a workers movement isn't one of them.

Technically they did kill the land owners and hand the land over to the workers. Then they murdered countless intellectuals and elites to make sure that workers had no competition in being on top. Then they enslaved the workers, tortured them, stole their production and took back their land. Even today they largely keep their wages suppressed.

You could argue that core Marxist principles are taking a back seat to national strategy, but the CCP higher ups are definite believers of Marx even if it might be easier to describe their system as some kind of neo-mercantilism. To them the reality is the same as it was under the Soviet Union. They cannot achieve their goal of communism in a world where capitalism also exists.

It's strange though how theoretically communism should be inundated with empathy, but in practice it barely cares about people and treats their lives as a commodity. It would be hilarious how stupidly bad their logic is in thinking that a brutal philosophy would be allowed to expand and control the Earth, if it wasn't so sad. They like to think of themselves as the clever deceptive geniuses who know better and are truly logical, erasing religion and going all in on science, yet they do the most insane illogical things that guarantee the most basic kinds of failure.

Failing to establish the fundamentals of a state to give most workers a decent living is bad enough, but to do that in exchange for a failed attempt to control the world is even worse. Failure has limited consequences too, because they can simply lie about how things are so the normal pressures that should exist for them to do a better job aren't as potent.

> The Taiwan Strait is notoriously treacherous. Taiwan has few beachheads, tons of tall mountains, dense urban environments, and flat, open farm land. It would be a nightmare of casualties and would destroy Taiwan's economy for at least a generation. China doesn't want to suffer any of that.

That doesn't change what I said. They still don't need all of that. China is also of the Russian mindset when it comes to human life. They don't mind throwing a bunch of their people at a problem to go die, so casualties is not a big issue if it came to that. China has also shown itself very capable of taking negative results in order to reject something else. So if there is something they want enough, they can likely suffer for it.

> Other countries had sanctioned Maduro and members of his regime, but only the Trump admin charged him with crimes.

I mean, technically these countries don't formally publish charges for someone in an uncooperative foreign country, because the UN typically doesn't like going into a country to capture someone. That means they rely on the person coming to them or being extradited, but both of those require the person coming to their country so you can arrest them or avoiding them fleeing before an extradition. So you have sealed warrants that nobody knows about until the moment the person enters your country and they get arrested. The ICC was investigating Maduro and Venezuela, but their investigations have often been delayed and if they did have a sealed warrant for Maduro, nobody knows.

The way the US accelerated the timeline is that the ICC does not handle terrorism cases. By identifying that Maduro was assisting cartels that are on the formally designated Foreign Terrorist Organization list and that he was promoting the distribution of drugs into the US for _political_ reasons and these drugs have caused hundreds of thousands of deaths, there is a kind of justification. For example, how we attacked terrorists in some countries without permission from a country is by claiming they are Unwilling or Unable to sort it out themselves or assist.

Another important distinction is that, if Maduro had been rightly elected he might have had more protection as a head of state, but instead since he stole the election and this was internationally recognized, he didn't have the same kinds of protections.

Some people argue that less fentanyl comes out of Venezuela than other countries, but they ignore the fact that cocaine alone has killed hundreds of thousands of people in the US so it is its own issue.

Obviously there is some gymnastics going on, but we would not have done this if it wasn't important.

> Here's a quote from Trump: “Now I love Venezuela. You know, I love it. We’re doing great in Venezuela. We’re taking a lot of oil from Venezuela, and we’re getting along great with them. Billions and billions of barrels of oil is coming out of Venezuela. … It’s the old-fashioned way, right? It’s the old-fashioned way. To the victor belong the spoils, right?”

That's mostly Trump trolling. Venezuela is still a sovereign state as per the UN, but we're managing their transition period since they don't have a legitimate government. The US used to have an oil deal with Venezuela that was pretty standard and quite the international norm. Communists then lied to the people and told them we were stealing their oil. They got elected and nationalized the oil, so oil companies left and stopped investing. In the 1990s and 2000s, Venezuela tried to return to the old deal to attract oil companies back, but nobody wanted to come back due to the risk they they would just nationalize their investments again. We're not simply taking their oil as "spoils", we're paying for it and the deal is structured to reduce the perceived risk that investors face when they invest in Venezuela. The deal is worse than what Venezuela used to have, but that's expected when you betray the trust of companies and investors which now means there's a risk premium included.

> This is pure conspiracy theory. Whatever algorithm you're in has broken your brain.

Algorithm? It's called Reuters. It's called AP News.

> Mostly agree, but this was uniquely Trump and wildly unlawful. You can't just invade other countries and abduct their leaders. You also can't just blow up random noncombatants in the ocean.

Venezuela was responsible for countless thousands of American deaths and has engaged in wildly unlawful things, disregarding international law blatantly. Maduro was not the leader of Venezuela, so we didn't abduct the leader. We are definitely blowing up drug boats, though and have taken out almost 70 of them now. In the middle east, we've launched strikes on lots of unarmed people that we designated as terrorists and these are just people working for the cartels that have been designated as terrorists. They don't have to be carrying a gun. These groups have killed over 1 million Americans over the past 20 years, which is more than any terrorist attack. I think a lot of people simply don't understand the scale of the problem we're facing in terms of grayzone warfare. The US is uniquely targeted, because we've been the only country standing in the way of countries like China or Russia having their way.

I've been doing my own research and taking notes for maybe 15 years. I'm not sure there's any author that thinks the way I do about it.

If I was to point at the absolute fundamentals, I'd say the open trade in the 1800s that caused many countries to rely on external supply chains leading to mass urbanization and hollowing out their agriculture created countries that were far less self sufficient.

With millions migrating away from the farms and working the land, moving to the cities, many countries were less self sufficient and it created an extreme imbalance and vulnerabilities. World War I might not have happened without that.

The economic issues combined with World War I was ripe ground for social movements which in the past were not always existential. When Russia was taken over by communists, they began to weaponize communism around the world. Any country that was undergoing hardship was a prime target.

Japan and Germany both suffered greatly from the 1800s mass urbanization with a weakened agricultural sector and when trade began to shut down in the 1930s to protect local economies out of fear not doing so would create an opportunity for the communists, this crushed Japan and Germany.

It would be fair to say then that there was some wisdom to globalism, but there was also some ignorance and it was not well managed. Yet, had communism not been the threat it was, global trade may not have receded as aggressively as it did.

So Russian weaponized communism, in full understanding of what communism actually is and how it operates in brutal destructive ways to destroy a society, combined with international imbalances caused by poorly managed trade in a rapidly industrializing world is a pretty good summary of what I see as the real causes of World War 2.

I'm not accurate about everything and I've made mistakes even in this thread which I didn't notice until after the edit option was gone, but I do make an effort to identify mistakes when I find an inaccuracy. Broadly, I think all the critical things I've said here will hold up as I haven't found convincing counter-arguments. If high quality counter-arguments are presented, I'm not closed minded and will consider them.