(untitled)

Ironically enough, AI chatbots like ChatGPT or Perplexity are now, in my opinion, better search engines than something like Google, whose stated mission is to "organize the world's information." No ads, no all-top-results-are-sponsored-links, no location-based results. You just ask the copyright-infringing botnet a question and it gives you a straighforward, no-bullshit answer

Not ironically, that is the entire point. Once most of the internet is unsearchable by humans, anyone who wants an answer to something will just have to trust the all powerful AI librarian to give answers based on a bunch of hidden information it has trawled. The capacity for a human to search for the actual source of that information will be phased out. I have been given answers by Google's AI answer bot that were diametrical opposites in search queries 10 minutes apart. If such an AI starts falsifying history, and no one bothers to look for original sources anymore that can be human verified (or can't because those results aren't shown) , then the entire past of the planet will be changeable over short periods to whoever controls/trains the AI. I have noticed the exact same thing as you have, but I don't think it is good. The search engines have, I think, deliberately allowed the quality of their ordinary search to worsen while the AI continues to have access to all the links that aren't prioritized for humans. Humans are just ad-cattle, and the search engines are there to direct human eyeballs to products. The AI is a completely opaque replacement for the original, research oriented value of the internet.

It's just a way to serve you up more ads on site in the event Google gets broken up separated from it's ad services. I think Google becoming bad at results wasn't really intentionally just they likely don't know how their own algorithm works anymore, tweaked a few things to increase ad revenue and it ate itself paired with their guidelines for web design that soykaftify the internet. Now even small search engines are better than Google for certain searches.

>>81826 By design imo, search engines have purposefully been ruined to promote the major corporate websites (reddit, insta, youtube, even stackexchange) or useless SEO AIslop. >>81839 Honestly I'm beginning to think webrings will make a soft comeback if search engines become continue to become more unusable; maybe a search engine could be built from a list of curated webrings?

>>81839 >the entire past of the planet will be changeable over short periods to whoever controls/trains the AI. Precisely. This is their plan. They don't want free information exchange because it threatens their capacity to conceal the illusion of their power

I appreciate that the OP pic is of the AI being dog soykaf retarded.

Shit got smart sadly. Also modern search engines are dismal, funny thing is, chatgpt isn't any better. Ever tried talking about something that isn't English? Something obscure? Free access to information doesn't exist at this point. Sadly.

fuarrrking retard of a vichan lost my screenshot. fuarrrk you!

>>81826 >Perplexity Are you retarded or what? All search engines have AI now even duckduckgo

>>81863 Instead of webrings I think curated search engines like Wiby and Mwmbl would be more likely to be more popular but webrings may come back to a mild degree

>>>>81934 OP is not retarded. Perplexity (and Phind) have been around since before ChatGPT. About a year ago they were the first guys who could correctly answer "How can I make a NAND gate out of only XOR gates" (some base models still can't get this) and they seem pretty resilient to hallucinating.

It's stupid Okay, to make the 1/s^2 transfer function steady and able to correct error, you can add an integrator in the forward path. This will introduce a pole at the origin, which will allow the system to have zero steady-state error. The modified transfer function would be: $$\frac{1}{s(s^2)}$$ This can be implemented by cascading an integrator (1/s) with the original 1/s^2 transfer function. The resulting system will have the following properties: 1. Steady-state: The addition of the integrator (1/s) ensures that the system has zero steady-state error, meaning it can correct any constant input error. 2. Stability: The system is now stable, as the pole at the origin is canceled by the zero at the origin introduced by the integrator. 3. Transient response: The transient response of the system will be slower compared to the original 1/s^2 transfer function, as the addition of the integrator introduces an extra pole. The trade-off is that the system will have a slower response, but it will be able to correct any constant input error. The specific transient response characteristics can be further tuned by adding additional poles and zeros as needed.

anything with the slightest amount of semantic comprehension is better than a search engine, even >muh google

>>81967 actually let me correct myself: better than any _existing_ search engine. google was decent before it decided to quantize every word to a set of normal user topics like {justin beber, covid, recipe, howto tie shoes}

>>81934 >All search engines have AI Not on my machine.

how can i legally access a fully unrestricted ai chatbot ingame in order to obtain completely legal and ethical information? i'm not familiar with coding or any deep tech stuff(USER WAS WARNED FOR THIS POST, You should always elaborate on your opinions rather than just spilling the soykaf.)

download gpt4all, download an uncensored gguf model, profit https://huggingface.co/bartowski/Lexi-Llama-3-8B-Uncensored-GGUF/tree/main here is an example as a general rule of thumb the higher the filesize the greater the requirements to run

The training step is learning how to read. Then you read some books. If you can't find what you're looking for in any publicly available document the AI won't know either since it's trained on the same info.

>>81826 I agree with this a lot. I've always been frustrated by having to change search engines to find what I want. I have to switch to google to find images, duckduckgo to find less politically censored results, and yandex to find copyrighted material. Having a machine that can find all this for me without doing all the switcharoo nonsense is really rather helpful in my opinion.

I don't think search engines failed because they were intentionally undermined. It's just that businesses figured out there's money to be made by having a high ranked website. And when millions of businesses are willing to invest millions of dollars into getting worthless soykaf onto page one, not even google can do much, even if they wanted to. Meanwhile, making a good and informative website is quite expensive, and if your only income is ads then it's not even financially sustainable. People did it for passion at first, burning money in fact. The free market optimized this early web out of existence. Chatgpt is both outstanding and awful as a search engine, but I don't think it's a good replacement

>>83183 this and when everyone started using google it no longer became necessary to put so many links everywhere to help people find interesting sites while web surfing. Such links which were necessary for google to find new sites and rank them. This and a lot of stuff moving into apps and walled garden type sites.

>>81934 If you use mainstream search engines instead of metasearch ones you're the one who's retarded m8.

Training is easy but you don't need to do it in most cases. A RAG setup is simpler and can have better results, it also takes less time to setup.