Look up “dataloom”. People have been playing with this idea for a while. It doesn’t really help with spotting errors because they aren’t due to a single token (unless the answer is exactly one token) and often you need to reason across low probability tokens to eventually reach the right answer.
Check your thinking. Korea currently has a DMZ dividing it from a war that never really ended and was fought to a stalemate. Their nuclear program didn’t result in military action because they currently have a gun to the head of every South Korean citizen and the backing of a large nuclear neighbour. Those are circumstances you can’t easily recreate elsewhere.
Adding to your point, Seoul is visible from North Korea, and vice-versa, and likely has enough conventional artillery aimed at it that even without nukes an invasion would be Very Bad for the Korean people.
North Korea is in such poor shape that they probably can't maintain much of the equipment much less keep the personnel trained and ready to use it effectively. Not a reason to go to war, but the threat to Soul and SK in general is likely massively overstated.
I think the strategic rational for unification completely swapped about 20 years ago. Up until the early 2000s it was likely in South Korea's, and the US's, interest to find a way to topple NK and unify the peninsula. The two populations had blood ties and common culture. Technologically the gap was growing but still reasonable. It would have been close to an east/west Germany type of situation where unification took effort but ultimately was clearly beneficial. China (and Russia) would have been losers in that unification would have brought a western friendly government even closer to their border. Additionally, NK still had a chance of re-energizing and becoming a real threat to SK.
Now however NK is in such bad shape that unification would be traumatic. South Korea would take on a problem of epic proportions, caring for and bringing a population of that size back into the broader world would be exceptionally costly and definitely not guaranteed to end well, possibly destabilizing SK in the process. Their cultures have grown apart making it hard for them to understand each other. The blood ties are not really there anymore. China and Russia would likely be the winners in that everyone sees NK as crazy and anyone helping them is hurting the world so they could get rid of that baggage. China especially would gain by having rail access to massive shipping assets to deliver goods even cheaper to the world. Finally, the US would loose a major rationale for stationing forces that close to China. They could, rightfully, say that NK isn't a threat and the massive US assets in South Korea and Japan should be drawn down.
It takes more than stockpiles of shells to be able to use it and maintain offensive positions capable of causing harm. From the reports I have seen NK military in Ukraine has been mostly cannon fodder and they are very untrained. That being said, joining the war effort in Ukraine is likely increasing their readiness.
Right... shells age. They blow up in the barrel, things like that. Maybe they even intentionally blow up in the barrel. Not that I would suggest sabotage. There's no way South Korean intelligence could possibly infiltrate North Korea ;)
But even so, if there was a serious threat of war, which is unlikely because China would stop North Korea, the US would place assets in the region and as we got close to a confrontation the US and South Korea (and as things are looking, probably Japan) would begin an aerial and missile bombardment to destroy in place North Korean offensive capabilities. Some would get through of course, perhaps thousands or tens of thousands of South Korean casualties, but in the context of a conventional war North Korea's capabilities would be quickly overwhelmed, at least in my opinion.
But honestly, the current status quo works pretty well for everyone except the people of North Korea, but there's not much we can do. It's a tragedy and the blame for that falls squarely on the Soviet Union and Chinese Communist Party.
... And the US, who razed every building in North Korea and killed more than 10% of the entire population of North Korea (that's entire population, including civilians).
Nope. US was there under a UN banner, and the UN force was winning until China threw manpower into the war. Never mind Soviet support. The blame goes to the communists and them alone. Without them Korea would have been likely unified under what is now the democratic South Korea we know today, but the communists in China couldn’t have a democracy so close to them, so they fought to win and establish the brutal regime that we have today in North Korea.
The only way unification can ever happen will be with Chinese blessing, with or without democracy. That would mean a full exit of US forces from the peninsula, and substantial pandering to the CCP and influence in Seoul. Which isn't that far off a thought honestly - for the most part, Korea was a tributary of China. With rapidly changing demographics and economic heft in both countries, it's even more likely SK will gravitate towards China, to the point that the Chinese will find more persuasion in unification and predictability.
All probably close to correct. I wasn't arguing that unification would, or should, happen (especially by force). I was arguing that the strategic value to China, SK, the west, etc have flipped as well as the actual capabilities of NK are likely vastly overstated.
The joke here is that in 20-50 years I'll bet worst Korea will be begging best Korea to reunify given the absolutely apocalyptic state of TFR south of the 38th parallel.
None of that would stop the current US administration from launching a sneak attack as we've seen several times in other countries. They simply do not care about consequences.
1. This sounds more like influencer marketing, I think people are already sick of it.
2. Yes and no. Depending on how you train the model they can output things that you’ve never seen before but the question is whether you want to look at those things. So yes a human has to judge and fine tune the output. This is why many models seem unoriginal, they’re designed to emulate specific styles and tuned based on broad appeal. If you go looking for LoRAs and merges created by “artists” you will see shit you couldn’t dream of.
Perceptible latency is somewhere between 10 and 100ms. Even if an LLM was hosted in every aws region in the world, latency would likely be annoying if you were expecting near-realtime responses (for example, if you were using an llm as autocomplete while typing). If, say, apple had an LLM on a chip any app could use some SDK to access, it could feasibly unlock a whole bunch of usecases that would be impractical with a network call.
Also, offline access is still a necessity for many usecases. If you have something like an autocomplete feature that stops working when you're on the subway, the change in UX between offline and online makes the feature more disruptive than helpful.
This is simply yet another outdated analogy from haters that are failing to keep pace with the current frontier because they are too busy getting high on the anti-hype.
We’re well past the need to retry the same prompt multiple times in order to get working code. The models with their harnesses are properly agentic now, they can find the right context, make a plan, write the code, run the tests and fix the bugs with little to no intervention from a human.
The hardest part now is keeping up with them when it comes to approving the deliverables and updating the architecture and spec as new things are discovered by using the software. Not new bugs but corrections to your own assumptions you had before the feature was built.
> This is simply yet another outdated analogy from haters that are failing to keep pace with the current frontier because they are too busy getting high on the anti-hype.
Try writing more documentation. If your project is bigger than a one man team then you need it anyways and with LLM coding you effectively have an infinite man team.
But that doesn't actually work for my use cases though, plenty of other people have already told me "I'm Holding It Wrong" without actual suggestions that work I've started ignoring them. At this stage I just assume many people work in very different sectors, and some see the "great benefits" often proselytized on the internet. And other areas don't see that. Systems programming, where I work, seems to be a poor fit - possibly due to relatively lack of content in the training corpus, perhaps due to company internal styles and APIs meaning lots of the context is taken up simply detailing takes a huge amount of the context leaving little for further corrections or details, or some other failure modes.
We have lots of documentation. Arguably too much - it quickly fills much of the claude opus context window with relevant documentation alone, and even then repeatedly outputs things directly counter to the documentation it just ingested.
So how are you as a person able to keep all of those rules in mind when you make a change? How would you train a junior engineer to do your job? Perhaps looking at it from that angle will solve your problem.
About religion, I don't think we can say "always" or anything near to that.
I agree that religions commonly use the god/god's will as the reason, but I don't think we should take that at face value. It's the argument to trump all others - rulers often claim to be chosen by the will of the supernatural - but not the reason the rule was made, which is a product of the cultures involved.
And humans often come to the same ethical conclusions: The rules against murder and rape, the priority on justice and fairness, as examples, are universal across cultures regardless of religion (look up 'cultural univerals').
Only if you are paying per token on the API. If you are paying a fixed monthly fee then they lose money when you need to burn more tokens and they lose customers when you can’t solve your problems within that month and max out your session limits and end up with idle time which you use to check if the other providers have caught up or surpassed your current favourite.
reply