LLMs draw origins from, both n-gram language models (ca. 1990s) and neural networks and deep learning (ca. 2000). So we've only had really good ones maybe 6-8 years or so, but the roots of the study go back 30 years at least.
Psychiatry, psychology, and neurology on the other hand, are really only roughly 150 years old. Before that, there wasn't enough information about the human body to be able to study it, let alone the resources or biochemical knowledge necessary to be able to understand it or do much of anything with it.
So, sure, we've studied it longer. But only 5 times longer. And, I mean, we've studied language, geometry, and reasoning for literally thousands of years. Markov chains are like 120 years old, so older than computer science, and you need those to make an LLM.
And if you think we went down some dead-end directions with language models in the last 30 years, boy, have I got some bad news for you about how badly we botched psychiatry, psychology, and neurology!
You are still talking about low level infrastructure. This is like studying neurons only from a cellular biology perspective and then trying to understand language acquisition in children. It is very clear from recent literature that the emergent structure and behavior of LLMs is absolutely a new research field.
Embedding „meaning“ in vector spaces goes back to 1950s structuralist linguistics and early information retrieval research, there is a nice overview in the draft for the 3rd edition of speech and language processing https://web.stanford.edu/~jurafsky/slp3/5.pdf
That's because .Net 4 has been the .Net Framework's current version since 2010. It's basically the same reason they never made Windows 9.
They dropped the Core designation because they're still trying to encourage people to migrate so they can take .Net Framework out behind the shed where Silverlight went. v5 was a convenient time to start that whole process of re-integration.
That's because they primarily sent stealth aircraft and Tomahawks over Baghdad. They also used decoys to draw out SAM missiles, and then F-4s would strike the SAM sites directly, which over time meant that the surviving SAM launchers did not fire when targets made themselves known. However, they did do some non-stealth missions. The most well know was Package Q, which involved dozens of aircraft, and two F-16s were shot down.
The thing about the First Gulf War was that it was four months of buildup, 45 days achieving air superiority, and about 100 hours of a ground war. It was well planned, and involved a collation of of forces that shared a common purpose and common goal. The allied coalition made sure to get their intelligence correct and worked hard to disassemble the Iraqi defenses before sending the armed forces into real danger.
The current conflict involved Donald Trump thinking that Iran, a nation of 93 million people with a relatively healthy economy (at least at the national and regime level, which can sell a lot of petroleum), was going to put up the same kind of fight that Iraq did, then a nation of 18 million with old tech, or like Venezuela did, a nation of perhaps 30 million today, that has faced extended total economic collapse, hyper inflation, and a mass exodus of something like a quarter of the population over the past 6-10 years. There was virtually no planning, with initial action going off of intelligence of where Khomeini would be and just jumping at that.
We've got an administration run by a narcissist that has surrounded himself with sycophants and bottom feeders. He's pissed off every ally we have, acted prematurely as the aggressor with an assassination strike, and now doesn't have the resources to protect the strategic assets in the region let alone convince Iran that the conflict needs to end in our favor. Just a ridiculous number of unforced errors. A complete embarrassment.
NB: The Palantir were created by the Elves, not by Morgoth or Sauron. The problem is that it takes a lot of will to use one and not have things of importance hidden (it shows what you think is important, not what is important), and as it turns out holders of one stone can influence what holders of other stones can see, if their will is greater. The Enemy doesn't get ahold of a stone until Minas Ithil falls and becomes Minas Morgul, and that's well into the Third Age. Two thousand years after the Last Alliance of Men and Elves, the second defeat of The Enemy, and the first destruction of Sauron. Which is still a thousand years before the start of Frodo's adventure. Lots of time in Middle Earth.
The rest of your comment is, unfortunately, spot on.
It's amazing how some people can read Tolkien's works and come away with an idea that it would be a good thing to create new powerful artifacts that will inevitably fall under the control of evil. Perhaps because their minds have already been corrupted by the One Ring.
I suppose it's just the Don't make the Torment Nexus effect with a different motif.
Eh, almost every link on the homepage has an initialism or acronym in the title, and roughly none of them are actually defining the term they're using. Indeed, not to point fingers, your own submissions make the same mistake.
Sure, yes, OP should (and now has) defined the term. But at the same time it's reasonable to expect that someone reading a blog post on BoringSQL.com would already know the term just as much as we could expect people interested in Clojure would know what a REPL is.
Really, you've always had to be careful with MySQL. It really was the PHP of RDBMSs.
The silent "SHOW WARNINGS" system, nonsense dates like Feb 31, implicit type conversions like converting strings to 0s, non-deterministic group by enabled by default, index corruption, etc.
That's true, but it might not be as important here.
Spain is not a country with a Common Law legal system entirely like the US or the UK. They have a civil law system where prior court judgement does not form a strictly binding precedent. Prior judgements can be important, but case law is not really a thing.
I wonder how true this is, we have the same system in Sweden, that court judgement are not legally binding precedent for lower courts. But in practice lower courts will follow the rulings made by the high court.
It's the same in Spain, which makes OPs proposal kind of useless. The big distinction between a civil and a common law system is the fundamentals. A country's civil code is properly defined, while a common law's system is based on previous cases you have to dig through to find the basics.
It's an extra cost. $100 to $200. You can't buy it, generally, except through a volume licensing partner. You may need to have a tenant ID depending on exactly how you're getting it, too. Alternately, you need to have a Visual Studio subscription which is $3k/yr. Oh, and you can't upgrade to LTSC from Pro. You have to do a fresh install. And IoT LTSC is even worse.
No, the problem with this is that a lot of IPs aren't profitable in their initial years, and this pretty strictly encourages property-holding as a business. That's exactly the wrong kind of revenue generation that copyright is supposed to be encouraging. It's empty rent-seeking.
Further, I think that the premise is flawed. Rather than being more protected by being profitable, a work should be less protected the more it has profited the owners. If you can make $50 million profit as an individual from your creative work that took 5 years to produce, then you're done. Dozens of lifetimes of wealth for 5 years of work? No, that's more than enough. You don't deserve more money for that. You have been suitably encouraged. The trouble with that idea is that "creative accounting" is too easy, so that won't really work, either.
I think it should match patent law. 20 years, and that's it. After that, if you want to keep making profit, you need to make something new. Because that's what it's supposed to do: let you make a living if you're able, and encourage you to keep working to create more.
LLMs draw origins from, both n-gram language models (ca. 1990s) and neural networks and deep learning (ca. 2000). So we've only had really good ones maybe 6-8 years or so, but the roots of the study go back 30 years at least.
Psychiatry, psychology, and neurology on the other hand, are really only roughly 150 years old. Before that, there wasn't enough information about the human body to be able to study it, let alone the resources or biochemical knowledge necessary to be able to understand it or do much of anything with it.
So, sure, we've studied it longer. But only 5 times longer. And, I mean, we've studied language, geometry, and reasoning for literally thousands of years. Markov chains are like 120 years old, so older than computer science, and you need those to make an LLM.
And if you think we went down some dead-end directions with language models in the last 30 years, boy, have I got some bad news for you about how badly we botched psychiatry, psychology, and neurology!
reply