DDG has a noAI portal that filters out AI images and doesn’t bother you with summations and things. it’s available at noai.duckduckgo.com and you can add it as a separate search engine to Firefox thusly.
I used DDG last night to search something with the hide ai images option checked and third result was from aiwallpapers.com
Yeah I know it’s a new feature and there’s gonna be some kinks but come on IT’S IN THE URL
there’s also a yesai.duckduckgo.com version, lol
Don’t know why you would, its as good as any other ai search summary and completely private.
Can I combine this with start.duckduckgo.com ?
And is there a mist of all the *.duckduckgo links?
Yes you can, by using the settings: https://start.duckduckgo.com/settings#aifeatures
And you can use the three dots menu on each link in the search results to file that a result is AI slop or otherwise not trustworthy and also filter domains from your future search results.
Fan of DDG, but i find SearX with better results
thanky
I don’t mind when it links to wiki but I actually follow the link to verify. It totally made up a pardon from the king of England to some guy yesterday
I’m going to say something spicy here, but for me personally, I’ve found DuckDuckGo’s AI search summaries to be quite useful. Not for the actual AI summary text, but for the links they give, which are often better than the normal search results.
That being said, I could easily do without them.
A spicier take still: I personally have found DDG’s AI summaries useful even without further clicking. When one’s query is purely technical (vs politics or whatever), I don’t see any need to click dutifully.
Same for me with Perplexity.
Me too, but fair warning to double check on the links if it is something niche. Perplexity can not always be trusted to interpret limited information properly. It does a pretty good job on enough for me to use it though.
As I said elsewhere, the problem is in fact that search engine providers deliberately make their search results worse to push AI usage. This keeps the user entirely under their control and at the same time hurts the websites the AI training data was stolen from, because no one will bother to visit them any more. I’m not saying DDG does this, but they get their search results from other search engines where this is the case.
Is that a documented fact that they make old search worse to promote AI?
The fact that search engine results gotten worse itself and that this was done deliberately is well documented, and it is documented that Google and others have a history of trying to prevent users from clicking through to the actual websites and keeping them in their ecosystem. They have developed similar things in the past, like Google AMP.
I have no definitive proof that they worsen their search results for promoting AI, but if you look at this thing there are a lot of indicators for this to be true. Controlling what the user will see and where they will go next is vital for these companies and it’s the reason why content algorithms exist and why they are creating “bubbles” to put individual users into. It’s all about controlling the content the user will see. Now if you think about it and ask yourself if having an AI box dominating the upper half of the screen giving you answers that the search results below don’t is beneficial to these goals, the answer is most likely yes.
Also you can do your own experiments which will make it pretty evident. Search for a few more obscure search terms. Use niche topics that will not yield a lot of results. In most cases the AI will nail it and the search results below won’t. Even if you use advanced search techniques it is really difficult to get the information that the AI gave you as a regular search result. But when you ask the AI for a source you get a website which has the content you were looking for.
Now the question is: Why is the AI that much better than the regular search engine? If you have used Google in the past, only a few years ago, it was perfectly possible to get those results through regular search, which is now bordering on being impossible. Odd, isn’t it? It seems like they gave AI a much bigger index to work with than their own search engine.
I have been wondering wether this is the case too. The search results on Google have really worsened the last few years, in my experience.
IMO that is a case of an unintended but welcome outcome for those companies
Google for sure did, you can read about it here: https://www.wheresyoured.at/the-men-who-killed-google/
Idk if DDG did similar or if they did, if it is documented.
Note that, reading the article & a recent follow up, it was moreso serving more ads that drove them to make results worse, rather than AI: the article was published in 2024, and refers to events starting in 2019. GPT2 got released around that time, way before ChatGPT (2022).
Still 100% enshittification though.
Wow that’s a really insightful article, I doubt a lot of people know about this, Thanks!
While we’re at it, I also like that they give me an AI chat that is ostensibly more private than alternatives for the times it’s useful. And choosing different models is great.
I agree. I don’t trust the AI at all, but the links are quick and easy to use.
You can also set them to only show up when you click a button for them, which I always preferred.
Unfortunately not possible when using temporary containers
Temporary Containers Plus is a Firefox extension that puts all containerless tabs in temporary throwaway containers that get deleted soon after they become unused
It does, however, interfere with saving site settings because cookies won’t be saved.
This is perfect for my use case. I mostly think AI results are a waste of energy, but having them on demand can be useful.
This can also be achieved just by changing DuckDuckGo’s settings using the menu in the top-right corner of the page and can turn off other things including adverts if you want to.
That works as long as you have the cookies for it; it won’t work in private browsing. Using OP’s method works in private browsing, too.
Both are good.
There’s a button “Show Bookmarklet and Settings Data” that saves all the settings to query parameters
Saving this for later. Thank you!
I remember when cruise control first became widespread for cars. Most people didn’t use it or barely used it. Some people, like me, did a lot of testing and figured out the best ways to use it, and ended up using it more than most. But then, there were people who just assumed it would work perfectly like they imagined, and used it as if it was a full-self-driving car, which immediately had bad results.
I think the worst thing about AI is that it lures people into fully trusting it, and they don’t even realize that their cruise control car is heading off-road towards a cliff. AI can be a useful tool if you know what you’re doing, but it is such a bad idea to have it on by default. Even a lot of fairly experienced users are tricked by AI. The average person doesn’t have a chance. It’s irresponsible to expose them to it.
I was going to mention about this whole thing with Winnebago and a driver assuming cruise control was FSD, but turns out that story is completely false. My father told me that story 10 years ago and I never thought to fact check it…
I don’t have a link, but I am sure I saw it on the news in the early or mid 90s. But one thing I have learned recently is that many of the “news” articles about cars are invented stories planted by other car companies.
Like one recent thing you’d have seen is stories about electric cars catching fire. It seemed that every time any electric car caught fire, it was national news, but non-electric cars catch fire frequently, as well.
So anyways, long story… less long, the story I’m remembering might have been fake, as well.
Somebody at work showed me keyword shortcuts ages ago and I have tons of them now.
Opera had this 20+ years ago. Once you get used to it, there’s no going back.
What are those?
Well I’m so glad you asked!!
You’re looking at one in the screenshot. Firefox does this, as does Chrome and some other browsers as well.
A bookmark keyword is a tiny bit of text that you can configure your browser to treat differently when you use it in the location bar.
Typically, whatever you type into the browser location bar will either treat that text like a website you’re trying to go to (like “apnews.com” or “ www.wikipedia.org ”) or text that gets sent to a search engine (like “tasty dinner ideas” or “best white socks”). However, if the text you enter starts with a bookmark keyword you’ve set up, the browser will insert the rest of the text you entered into a website address in a specified place.
This is typically useful to speed up searching on specific websites.
So if you want to search Wikipedia for “particle physics”, you can go to the Wikipedia website and enter “particle physics” into the search box and click the search button. That would send you to a page with search results of the text you entered. If you look at the location bar, you should see a URL that looks like this:
https://en.wikipedia.org/w/index.php?search=particle+physicsWhat we notice here is that the text you entered, “particle physics” is right there in the URL.
To turn this into a bookmark keyword, you create a bookmark to this search results page, then replace your search term with the characters “%s”, so the bookmark URL would look like so:
Then, in the “keyword” box, you can enter whatever text you want to use for this shortcut. For Wikipedia, I like using just the letter ‘w’. (You don’t need quotes around it.) Save the bookmark, and that’s it.
Now, whenever you want to search Wikipedia, all you have to do is type “w particle physics” or “w forest fires” or “w whatever” into the location bar and the browser will take you directly to the search page with those results.
You can do this with basically any website with search functionality: search engines, retail stores, news, IMDb, reference resources, whatever.
This feature also can be used for going to detail pages directly if you have a specific reference number.
So let’s say you’re at work and you have a trouble ticketing system that shows details of ongoing issues. The URL for ticket number q-rt-654321 might look like this:
https://troubletickets.mycompanyfoo.biz/ticket/q-rt-654321/viewSo if you had the ticket number handy (like from an email chain), you could create a bookmark keyword to go directly to the ticket detail page:
https://troubletickets.mycompanyfoo.biz/ticket/%s/view…and use the keyword “tt” for trouble ticket.
Now you can just type “tt q-rt-654321” into the location bar and go right to the detail page (presuming the ticket number is accurate).
And that’s it.
I already knew about using this for search, but the ticket example is another great use I didn’t think of at all! Totally using this, thanks.
I suppose this is faster than using !bangs on ddg
Cool thanks for the explanation, I’m off to make some bookmark keywords!
But it should really be “ai.duckduckgo.com” and off by default.
If you see AI you need to duck, duck again then go (as fast as you can)
The fact that something should be a specific way has absolutely nothing to to with the way it actually is.
F.W. Nietzsche, Morgenröthe
There is also the no-java site, idk if it filters out ai images, but doesn’t seem to have ai otherwise, no search assist.
Java is not Javascript. They’re not even related. Netscape chose the name to profit from Java’s poularity back in its’ day.
Thanks for the correction. The message does say “redirecting to non-javascript site” when I search with noscript installed.
there’s also https://lite.duckduckgo.com/
AAAA wtf im trying to post the URL parameter for easy copying, but it always gets converted to this
Where the fuck is that 25 coming from?
If i write /?q=%s seperately like this its not problem. This even happens on mobile, so its just not allowed to post this kind of link for some reason.
Oh without the https:// it doesnt happen
It’s getting encoded. % is a special character in URIs. Let me try posting it inside of back ticks, as well as triple back ticks:
https://noai.duckduckgo.com/?q=%25shttps://noai.duckduckgo.com/?q=%25sI’ve noticed that there’s a plague of threadiverse clients which improperly escape/encode URIs. It’s most evident with how they mangle parenthesis in Wikipedia article titles.
%20 is the URL-encoded form of a space; %25 is the URL-encoded form of the percent sign. The URL you are posting gets re-encoded and % becomes %25 (in the same way that a space becomes %20)
It’s encoding your URL into the correct format for your Browers URL bar. “%25” is the encoded version of “%”, so it’s being replaced.
It’s obviously incorrect in this instance since it’s not an actual URL. Maybe it wouldn’t happen if you’re able to put the URL string into a code block?
Why tf is http not unicode yet?
…or to Vivaldi, or vanilla Chrome, or any other browser that has search engine options in its Settings.
Even edge has that nowadays.
