Writer, Disaster, Charmander

BUT

cause like. apart from the (potential) exception of writing code, there is not really anything useful that a Large Language Model can do.

it imitates patterns in the text that it takes in

that's fundamentally how it works

Yes, haha funny, the AI is bamboozled by spelling, but seriously: I asked Google the same question 5 different times just now, and not only did I get 5 incorrect answers, I got 5 completely different incorrect answers.

The fact that they're different every time demonstrates just how much of an LLM's output is up to random chance. This is how LLMs imitate human language: each stage of constructing the text has built-in randomness, just like how human language has variation and flexibility.

This is my problem with using these things in the workplace. Like, imagine you're using an LLM to fill out a report of an accident that happened, or to write dietary recommendations for a patient, or to give feedback on a paper a student wrote...really anything you can think of.

Even if the LLM-generated text doesn't say anything technically flat-out wrong, the words and details are still randomized instead of being chosen by you for an actual reason.

Sometime in the past ten years, consumer-grade technology has hit a point where it just...doesn't really need to be more powerful. In the tech industry (or at least its shareholders), this is a disaster because they have been accustomed to rendering everyone's tech completely obsolete every so often and then getting massive influxes of cash as people replace their old hardware. Back in 2006, a desktop PC from 2000 would be completely useless. Today, a high-end laptop from 2020 is not only serviceable, it's still high-end.

Nvidia's business model has been about releasing a brand new hyper-powerful graphics card every single year for like over two decades, and they don't want to change that, but like, 4K literally quadruples your power demands and it doesn't look THAT different from 1080p. even the gaming enthusiasts don't really need em, and the cards have kind of scary power demands anyways. like am i going to find out something interesting about my home's wiring after i plug this thing in?

Crypto kept the computational power arms race dream alive for a little while, but you can't actually do anything with crypto, and it's regulated by a lot of governments now so you can't even do crimes with it anymore, but hey, you know what else uses lots of fast linear algebra? Large Language Models.

While LLMs don't do anything well, they SEEM like they can plausibly do some things so long as you, the user, don't know anything about doing the thing, which is convenient because C-suite folks almost universally don't know how anything anywhere actually gets done, and tech investors tend to be excitable bc they want to invest in the next unexpected runaway success and all you really need to do is bluster at them a bit.

Nvidia is making a lot of promises and saying a lot of big numbers, and Ed Zitron has done a lot of fantastic research into how the economics(don't) work and what's actually getting built, and the short version is that none of this shit adds up. Personally, I'm pretty sure AI is a confidence game of some kind, but it's pretty obvious that AI is smoke and mirrors. The real question is: what happens when the music finally stops?

I can't know the exact future, but right now, LLMs are VERY heavily subsidized, and Microsoft has basically stated that their business model is just the Walmart scam, where they convince companies to fire all of their graphic designers, writers, and programmers by offering a cheap alternative, and then jacking up the price once the companies are reliant on it. the question here is whether or not a profitably-priced LLM is ACTUALLY cheaper than hiring the graphic designers back (i have my doubts).

I suspect that one of the other possibilities in play is that Nvidia and Microsoft are trying to build the infrastructure they'd need to force all of us into having to rent our entire operating system from them and owning only a remote terminal that we can use to access their servers. This is a very stupid idea that can't work for a lot of reasons (the main one is that lots of enterprise technology needs to function without doing that and your consumer tech market isn't wholly independent of enterprise tech), but that doesn't mean they're not going to try building it anyways, and I suspect it will actually mean that we all switch to using Linux and buying bootleg Chinese GPUs.

1. The court holds Google responsible for statements made by its AI, considering them Google's statements (search engines have limited liability for results in their engine as they're the words of other sites/companies/people), meaning when their AI lies/hallucinates they're liable for the defamation/harm resulting from those statements.

2. Google's defense that customers are generally aware of the lack of reliability and are responsible for fact checking was dismissed. As the court pointed out, that would "significantly diminish" AI Search's stated purpose and it can't be distinguished from Google's business practices/statements as a search tool.

3. Studies have found about 91% of Google's everyday AI responses are accurate, leaving millions of searches per HOUR with potential liability for falsehoods. 56% of correct responses weren't supported by the sources the AI listed. Both of which mean Google is now liable for a LOT more AI "errors."

4. Google was held liable for 80% of court costs in this case and this precedent is expected to reverberate around the world. This is a massive shift from the 3rd-party search provider role Google has previously played and it comes right as they've tied ALL searches to their AI search.

happy pride month to these two specifically, love whatever they have going on

But I know how to apply the gel because I saw posts from trans women how to apply it and I feel that even though are reasons for using it are different that we are not so different