翻訳待ち:Un-AI Your Internet
AI サービスが一時的に利用できないため、復旧後に翻訳を補完します。ソース概要:UN-AI YOUR INTERNET UN-AI YOUR INTERNET [ Click each section to begin. ] The State Of Things The scourge of LLMs have been shoved down our throats by our techbro overlords because it allows them to demonstrate that more…
AI サービスが一時的に利用できないため、復旧後に翻訳を補完します。
UN-AI YOUR INTERNET UN-AI YOUR INTERNET [ Click each section to begin. ] The State Of Things The scourge of LLMs have been shoved down our throats by our techbro overlords because it allows them to demonstrate that more and more people are using the things they have spent so much money investing in. Never mind that we aren't interested. Never mind that the thing they have labeled as "AI" isn't anything like what they are actually selling it as. The more people "using" the technology, the easier it is to sell the technology, even if in this case the word "using" is doing a lot of heavy lifting for "we bundled it with everything so it looks like more people are using it." "Agents" and "assistants" are the new buzzwords, but it's all the same thing: A complex algorithm that uses predictive text to piece the most likely, most used words together as a response to a prompt. The reason we haven't had these things until now is that processors and storage medium weren't powerful enough and cheap enough to brute force a response built from entire libraries of stolen data. This isn't a machine that thinks, its a decent simulation of what that might look like, using the power of ultra-fast processors generating "most likely" predictive responses. It's a scam. It's snake oil. And because its so ubiquitous, it's actually likely to get in the way of real AI development. It's the worst, most inefficient way of simulating a real AI. There are two paths: one path is the slow, laborious process of teaching a burgening intelligence how to think, how to process, and how to reason. We are a hundred years from that. The other path is to use predictive algorithms and reams of stolen data to inefficiently brute force a response that "seems like something an AI might say," and you can guess which path the techbros selected. The path that gives them results, sort of, right now, allowing them to ride the bubble until it bursts, hopefully long after they've retired to a tropical paradise somewhere, leaving us all relying on LLMs that frequently generate plausible sounding responses that are 100% wrong. I'll be putting "AI" in quotes whenever I use it, because I resent that they've short changed us with this nonsense masquerading as actual artificial intelligence. So What Can We Do? → You can, with some effort, remove or disable every instance of "AI" infecting your internet. It's a pain in the ass, but fortunately there are a number of tools available that make it a little easier. That is the main focus of this website. → You can, by doing some homework, stop supporting artists, writers, and developers who use "AI" in their work. → You can reject software that bundles "AI," if there is an alternative application. (There are fewer and fewer applications that are "AI" free.) The important thing is that this doesn't have to be something we just have to deal with from now on. Remember when everyone was very excited about NFT and blockchain technology for no reason? The difference is that every techbro and compute manufacturer has invested heavily in AI, so waiting out this particular bubble will be a lot harder, because they're going to be pushing it into your life to justify their investment. We're currently in a circular shitshow where AI hardware manufacturers, developers, and datacenters are just cycling the same money around and hoping nobody notices, and if they stop, the whole thing implodes, so it behooves them to never stop. While we wait, you can start blocking, uninstalling, and disabling the "AI" that comes pre-installed with your phone, tablet, laptop, and various applications you might use. Un-Fuck The Internet: The Steps This is a complicated subject, and it covers more than just AI. How do you want the internet to look? To operate? There are two distinct opposing thoughts here. The first is that the internet should be tightly controlled, to protect you and to ensure that you are seeing and experiencing exactly what the company that makes your computer or device wants you to see. This is part of the "walled garden" paradigm that is largely attributed to apple. The other path is for you to control how the internet looks when it's delivered to your computer or device. You determine how the experience goes. This is less safe for you, as a user, because you can't depend on the walls of the "garden" to protect you. You'll have to learn how to do that yourself. But it does allow you freedom that the users in the walled garden will not be able to enjoy. Here are the steps. We'll cover more detail on each section below. Step One: Stop Using Google Search. I know, that's very difficult for a lot of people. But the reality is that it's easier to use another search engine altogether than it is to repair all the "AI"-based damage that google has inflicted on its search engine. You can do it, with enough extensions, but why bother? Step Two: Stop Using Chrome. You will need to start using a different browser, because google is slowly (and sometimes not-so-slowly) killing the ability of its users to modify the way chrome works. They do this in the name of security, but the reality is, google makes money selling ads and your data, and extensions that keep them from doing those two things cannot be allowed to exist on their browser. So the second thing you need to do to un-AI your internet is to un-google your browser, just like you un-googled your search engine. This will allow us a lot more control over various aspects of your internet use. Step Three: Extensions. Specifically, we'll be installing either uBlock Origin or uBlacklist, two extensions which can be used to block websites. uBlock Origin allows us to also block website elements and other things throughout the internet, whereas uBlacklist only blocks sites from search results. Step Four: Block Lists. These are lists people have compiled of known "AI" websites and youtube channels and other things. Thanks to these lists being public, we don't have to block each of these sites one-by-one, we can just install an already made list. Step One: Stop Using Google Search Google has become synonymous with "search," but it didn't used to be that way. There were once hundreds of different search engines, and people used the ones that they liked more than the others, for various reasons like the UI or the types of results, or because of where and how it searched. Some of those engines are still around, but they've been bought and sold and absorbed and split off and redesigned so many times that it's hard to tell where they are pulling their results from at this point. However, there are modern search engines that are not google, and which I can certainly recommend. Indeed, you should switch to one of these immediately: DuckDuckGo Search, with the "noai" subdomain, is the search engine I would recommend for most users. It's free, it comes with several settings you can adjust to your liking (click on the 3 bars in the upper right and select "settings" or just go to noai.duckduckgo.com/settings), and it's fairly good at returning relevant results. Using the "noai" affix in the address automatically disables all of DDG's "AI" features, but you can also turn them off in the settings. It also has a setting for "Hide AI Generated Images" that is... relatively good at doing what it says. It really tries hard, but there is so much "AI" "Artwork" flooding the internet, that you can expect some of it to slip through. Though DDG states that it pulls results from several sources, it seems to rely heavily on the Bing index for a sizable portion of its results, based on the whole thing failing when there was a bing outage in 2024. Kagi (Kah-gee) is the next search engine I recommend, and it's probably the best non-google search available, at the moment, in terms of results and reach. Kagi is one of the only modern search engines that maintains its own index (so it doesn't rely on google or bing or yahoo for its results) and prioritizes non-commercial websites (though commercial websites do appear in the results as well) which leads to a more "small web" feel. The phrase "If you don't know what the product is, you are the product" (alternatively: "If you're not paying for the product, you are the product") is at the heart of Kagi, and they reiterate over and over again that there are no ads and they are not collecting or selling your information. This is great, but it also means you have to pay for a subscription to Kagi in order to use it. Kagi is not free. The cost is tiered, with $10/month being the cheapest plan with unlimited searches. Kagi does have "AI" features, but they are opt-in and can be ignored or disabled. Brave Search is a search engine from Brave Software, Inc. the company that makes the Brave browser (More on the browser later). Like Kagi, brave maintains its own index, though it uses google as a fallback when the index fails to return results. Unlike Kagi, brave search is free. This is where you wonder if you are the product, and indeed, you are. Brave sells ads, which are anonymous (unlike google's adsense ads) and conform to its own (fairly stringent) privacy practices, and which appear throughout its results. If you would prefer not to see ads at all (and still want to support brave search) you can subscribe for $3 a month or $30 a year, which removes all of brave's ads. However, you must use the brave browser for the ad-free subscription option. Brave is a good choice if you want a free search with its own index, and a robust privacy policy. Its results are largely in line with DDG. Brave has "AI" features, but they can be turned off easily in the settings. Qwant is a french search engine now used as the default by the UK parliament, as of 06/04/26, as the UK and greater Europe increasingly wean themselves off of google search. Qwant has its own indexing system like Kagi and Brave, which is the European Search Perspective or EUSP, a joint index created by Qwant, in France, and Ecosia, in Germany. Qwant is focused on privacy, does not track users, resell personal data, or bias the display of search results, according to their wikipedia entry. Qwant has some AI-related features, but they can be disabled. Ecosia is a German search engine thats main hook is being an eco-friendly search engine. It's powered by renewable energy sources and uses some portion of its profits (I am not sure what percentage exactly, but it seems to be about 1/4 of its total income?) on planting trees along with some other eco-friendly programs. That being said it does have a whole news post entitled " Our AI just got even better," the byline of which is: "We're building AI that is different. It answers to the planet, not Big Tech. And it's now more European - which means more independent, and more private." So... yeah. It also uses both google and bing for results in addition to the the EUSP index that it partners with Qwant for. So you may end up using google on the backend anyway. You can turn off AI overview in the settings. I guess using Ecosia comes down to how granola you are and how much you trust their eco-marketing. Technically, still not actually google, though. Some of the old engines still exist, sort of: Dogpile.com / Metacrawler.com / Webcrawler.com / startpage.com are all owned and operated by System1, an "internet advertising company." Upsides? None of them are google, and none of them have AI features (though startpage's settings includes links to vanish.ai, a private AI model). Downsides? It's an internet advertising company. All of these are "meta" searches, in that they use the indexes of multiple search engines, including google and bing. I think I would use DDG or Brave if you need a free search engine. Sti [truncated for AI cost control]