![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
And this guy balked at buying some when it was like 130 bucks
I don't see AI going anywhere anytime soon. It is a plague.
This is astronomically different than crypto. AI is permeating the entire world right now and NVIDIA seems to be putting themselves at the direct center of it.
They actually have for a long time. nVidia was pushing the "AI" bandwagon years ago when most of the industry didn't give a shit. DLSS is one example of a consumer tech that came out of that, it was real early to the game, and wasn't very good when it first debuted but they worked hard on it. The marketing back then was more "neural network" and "machine learning" and not so much AI as that hadn't become the buzzword but they've been pushing on it for a long time.
Yes, I know. Their foresight is paying off in droves.
I mean AI really isn't AI in the way that most people think of it. AI is just a marketing term. I'm not buying anything. I don't play the stock market. I feel a worldwide recession is highly probable at this point.
I am ignorant and I don't know what happened. Why did "AI" explode into news, marketing, investing, etc? I just never bothered to dig into it. Anyone care to chime in to educate a know-nothing?
ChatGPT. That's what did it. There are plenty of other important and useful developments in machine learning that have been happening for a long time, like DLSS/FSR, car assisted driving, etc, etc. However what made it news for the masses was GPT 3. Here was a chatbot that, to many untrained observers, seemed to be truly "intelligent" to actually pass the Turing test. That generated a huge amount of hype and news. Then, people started over applying it to everything. There were tons of "AI startups" that just slapped a wrapper on ChatGPT, companies laid off employees and replaced them with chatbots, often with quite bad repercussions, etc, etc.
In terms of pure animation though (not generative art) I've seen neural network models from a year and a half ago animate armatures automatically (albeit unpolished), so it just depends if the pace keeps up to where replacing humans like that in the near future is actually viable.
There was a major issue with system getting bigger that made the back error propagation or something of the sorts being such a small proportions of the total weights make them stop to work.
Do you mean the number of users, or what?
Yes, users after x first month of commercial existance, apparently:
You mean stackoverflow dotcom, the web site?
Yes:
Any stock professional will tell you that they can't really time the market. If they can't do it, what chance to us poor shlubs and wage slaves have of doing that?
That is pure, unadultarated s---, designed to hype the stock price and stoke demand for AI hardware. Someone I know who is a developer tried to get ChatGPT to write some programs. Terrible results. "Cliff Clavin" style of programming, he called it. https://en.wikipedia.org/wiki/Cliff_Clavin
depends what you mean at the current rate, no or in 30 years what the world can be, but look at the world internet traffic, it has been growing rapidly unstop since 1990
ChatGPT is not to program, in 15 years the amount of difference with the prehistoric early stuff of that genre, trying to code in language-system not made with their existance in mind. Often GPT trained on code without seeing the result of that code
Results vary wildly in terms of robustness but yeah. It's unfortunate, too, from the implication of 'stay uneducated' since our products will do these things for you. These things abstract entirely away the logic and thinking involved in such tasks that I'm not sure what the user learns besides knowing where to go for answers. Sure it's useful (when it works) but how is that competitive for the types of things that run the world, at least in the timeframe implied.
Assembly did that for machine language, than programming language-compiler for assembly, etc... it depends where we put the line and it will be quite subjective.
Just to clarify my position, I don't have any issue with people using such abstractions for getting something done, even down to just natural language prompts. It's just Jensen's quote is specifically addressing the types of people who would be considering studying computer science, ie: they've self-selected to learn about logic and the lower level workings of how computers operate to try and gain the best understanding for the field, yet he's saying that's (soon to be) unnecessary to be competitive to other programmers. That to me is the wild thing.
I feel you are perfectly right, but the kids that would be deterred by such comments is probably the one that will end up at high level anyway, the one that will learn not for a job in mind but because of a interest in learning will still be there.Just to clarify my position, I don't have any issue with people using such abstractions for getting something done, even down to just natural language prompts. It's just Jensen's quote is specifically addressing the types of people who would be considering studying computer science, ie: they've self-selected to learn about logic and the lower level workings of how computers operate to try and gain the best understanding for the field, yet he's saying that's (soon to be) unnecessary to be competitive to other programmers. That to me is the wild thing.
Or I'm just being too uncharitable with the interpretation
Hehe, I told everyone I knew to buy AMD when it dipped below $1. Myself, I don't purchase any tech stocks, because you know, conflicts of interest etc. And I am someone that never wants to get on the wrong side of the SEC.
YUP. SO is a great resource if you can find your question already answered, and if the technology you're asking about is relatively static. But heaven help you if you try to ask a new question, or if best practices have changed since a question was originally asked/answered. Sometimes you can find a good update in a comment on an answer explaining "this was correct, but don't do it like this any more"...sometimes you're left needing to know the specific technology well enough to make that discernment. I teach folks how to use Spring at work, and there's lots of SO posts filled with knowledge of how to do something with Spring a decade+ ago, when it was all configured with XML files...and getting good information on more modern practices is appreciably more difficult. Or answers that give you an answer on how to do something with Java circa Java 7 or 8...and disregard the newer stuff. Still a good resource, but has so much more potential. ChatGPT will produce crap much of the time, but at least it'll patiently sit there and let you ask your question over and over with any phrasing you want until you luck into something it "understands".
100 P/E is aggressive obviously but, up 265% in revenue, the evaluation is at 28.57x time the current pace annual gross profit, AMD is at 25x time with a 340 PE, Netflix at 17.9x, not that far away while being no near those growths.
ChatGPT can do trivial stuff pretty well. I asked it to create a tool that does the same thing one I wrote several years ago: I had 2-color bitmaps that had white borders around them that I needed to get rid of the border. ChatGPT could make a program that does that fairly well, but it couldn't handle the extra stuff I wanted: drag and drop multiple files, process them, rename the file from "whatever.bmp" to "whatever - original.bmp", and then save the new, cropped file with the original name. I haven't tried doing recently, though.
AI is not akin to crypto, but it is akin to the dot com boom. We still use the web and it's more prominent then ever. The same will happen with AI. Unlike the creation of the web which created a lot of jobs, the AI boom has the potential to take away a lot of jobs. Which creates a problematic economy where suddenly people are losing jobs due to AI, which reduces the purchase power of people. Who's going to buy these products when AI took their jobs? AI is making money now because right now people still have their jobs.