These numbers underline the current trend to choose European services instead of American ones, which followed the trend to deGoogle.
[the chart shows stats for American Google, American Bing, Russian Yandex, American Yahoo!, American DuckDuckGo, and Other]
JuvenoiaAgent@lemmy.ca
on 29 Apr 15:50
nextcollapse
Yeah, that statement wasn’t supported by the data at all. It seemed to only be included as a way to link to their other articles about European alternatives and de-Googling.
UnderpantsWeevil@lemmy.world
on 29 Apr 16:14
nextcollapse
Russian Yandex
Exceptionally good at finding torrent sites and other piracy outlets, because they aren’t working hand-in-glove with American broadcasters to censor and shadowban these links. Google, Bing, DDG, and the other American mainline search sites all focus on feeding end-users into a discrete set of Web2 mega-site sponsors. Yandex uses the older web crawlers and indexing tools, so it gives more honest (abet fuzzier and less reliable) results. And since nobody really gives a shit about Yandex, the efforts to game its algorithm have been comparatively minimal.
Yandex also has the benefit of being relatively English-friendly, while other popular non-English search sites like Baidu, Qwant, and Naver don’t cater too quite so freely.
Atelopus-zeteki@fedia.io
on 29 Apr 16:26
nextcollapse
Thanks for mentioning Yandex, bringing is back onto my radar.
And they have really good products - the Navigator is great, and Yandex Music was better than Spotify (until the war started and a lot of labels/artists disappeared).
I’m not using their products now as I don’t want to feed the government, but they do(did?) some great stuff.
Who still uses search engines to find torrent, though?
UnderpantsWeevil@lemmy.world
on 29 Apr 19:25
nextcollapse
Sites periodically get taken down or rendered less than useful. Especially for live streaming.
Yandex was invaluable when I was looking for Olympics streams, for instance. Also really depends on which communities are hosting to which torrent sites. I found nyaa.si off Yandex, because I couldn’t find the anime I was looking for on 1337x.to.
It’s been common ever since magnet links were created, since you can post a magnet link anywhere (even in a plain text file) rather than having to upload a .torrent file somewhere like in the old days.
mesamunefire@lemmy.world
on 29 Apr 15:49
nextcollapse
Google been degrading as time has gone on. The other search engines (like all of them) are getting or surpassing google in certain subjects. AI has really made them look like fools in all of this. Googles AI sucks for results and (while I dont like it) others are using chatgpt for search results.
I would say it is just the opposite. Google used to be good before they tried to post-process results in this extreme way and AI is just an even more extreme way to do that. ChatGPT and all the other LLMs just increase the noise to signal ratio (noise coming first because there is so much more of it than signal these days).
People use chat gpt to tell them what to buy. We are doomed. Our brains are about to shrink to the size of a pea in 10 years. All is going to plan for the elite.
Its funny because you can totally run good LLMs on local systems but people are just going to chat because its what they know and its easy to work with. Like I get it, but they are starting to put ads in the prompts now.
Google was best in the 2000s, but things were different back then. SEO spam wasn’t really a thing yet, there were far fewer websites, and most online discussions were archived and searchable (compared to today where there’s platforms like Discord that aren’t indexable in search engines at all).
JaymesRS@literature.cafe
on 29 Apr 15:48
nextcollapse
I work in an education setting and in the last month, Google started preloading the contents of other sites directly on the search page. It is wreaking havoc when combined with our blocking tools because kids will do a Google search for something innocuous and the page will immediately get blocked because it tried to load a result from Reddit or coursehero or something else we have blocked.
It’s incredibly frustrating.
not_woody_shaw@lemmy.world
on 29 Apr 16:15
nextcollapse
unwarlikeExtortion@lemmy.ml
on 29 Apr 18:54
nextcollapse
the page will immediately get blocked because it tried to load a result from Reddit or coursehero or something
Does that mean any search (AI insight notwithstanding) will get blocked if it includes a Reddit, Coursera or something on the blocklist result at all?
Because if yes, that’s much more than just asinine. It’s basically blocking entire search topics due to the sheer fact that Reddit will appear on the furst page of Google a lot.
That is exactly what is happening. They type in the search query on their Chromebook (for example, “why do dry erase markers float”, the results page flashes for a second and then the “this page is blocked” screen comes up saying they were blocked from Reddit, et al. Without them clicking on any search results.
And 2 kids can do the exact same search at the same time and get blocked for different sites or only one will get blocked.
We have a different person that manages our block list so I don’t know the reason for all the blocks. That said, we block www.coursehero.com not www.coursera.org
Oh I thought it was a typo. I have no idea what coursehero is.
I will say coursera is awesome for their calculus classes. (Or they were like 10+ years ago anyway.) Which caught me up after a long hiatus from college when I returned to finish.
To be fair, I don’t know that we don’t block coursera as well but the block I specifically saw was for a course hero. And some of the blocks that we have on are for security reasons over reliability reasons. We have to be hyper cautious about the students leaking any potentially PHI and some of the Google sign in setups are less secure than others.
Moobythegoldensock@lemm.ee
on 29 Apr 19:15
collapse
Google search has sucked for a couple years now. DuckDuckGo is better for everything except maps.
Routhinator@startrek.website
on 29 Apr 20:17
nextcollapse
Thats because for some ungodly reason they use Apple Maps. Not sure why they dont integrat with an OpenStreetMaps like service. At least that way users can start contributing to fill the gaps
That’s been my go to since they started. There is/was a challenge using them when we evaluated them a while back with forcing safe search reliability if I recall the reason.
Bishma@discuss.tchncs.de
on 29 Apr 15:54
nextcollapse
I went long enough without using Google (probably a year-ish) that, when I accidentally made a Google search a few days ago, it was a jarring experience.
It felt wrong the same way other search engines did when I first deGoogled. It was kind of nice actually.
brucethemoose@lemmy.world
on 29 Apr 15:55
nextcollapse
The irony is Gemini is really good (like significantly better than ChatGPT), and cheap for them (no GPUs needed), yet somehow they made it utterly unbearable in search.
Bishma@discuss.tchncs.de
on 29 Apr 16:06
nextcollapse
“Significantly better than ChatGPT” and “Good” aren’t the same. Like ipecac is significantly better to drink than sewage water.
Gemini is really good at confidently talking nonsense but other than that I don’t really see where you get the idea that it is good. Mind you, that isn’t much better with the other LLMs.
flamingo_pinyata@sopuli.xyz
on 29 Apr 16:11
nextcollapse
So it’s really good at the thing LLMs are good at. Don’t judge a fish by it’s ability to climb a tree etc…
I get the desire to say this, but I find them extremely helpful in my line of work. Literally everything they say needs to be validated, but so does Wikipedia and we all know that Wikipedia is extremely useful. It’s just another tool. But its a very useful tool if you know how to apply it.
But Wikipedia is basically correct 99% of the time on basic facts if you look at non-controversial topics where nobody has an incentive to manipulate it. LLMs meanwhile are lucky if 20% of what they see even has any relationship to reality. Not just complex facts either, if an LLM got wrong how many hands a human being has I wouldn’t be surprised.
LLMs with access to the internet are usually about as factually correct as their search results. If it searches someone’s blog, you’re right, the results will suck. But if you tell it to use higher quality resources, it returns better information. They’re good if you know how to use them. And they aren’t good enough to be replacing as many jobs as all these companies are hoping. LLMs are just going to speed up productivity. They need babysitting and validating. But they’re still an extremely useful tool that’s only going to get better and LLMs are here to stay.
That is the thing, they are not “only going to get better” because the training has hit a wall and the compute used will have to be reduced since they are losing money with every request currently.
Technology these days works in that they always lose money at the start. Its a really stupid feature of modern startups IMO. Get people dependent and they make money later. I don’t agree with it. I don’t really think oir entire economic system is viable though and that’s another conversation.
But LLMs have been improving exponentially. I was on board with everything you’re saying just a year ago about how they suck and they’re going to hit a wall even. But the don’t need more training data or the processing power. They have those and now they’re refining the LLMs. I have a local LLM on my computer that performs better than chat GPT did a year ago and it’s only a few GB. I run it on a shitty laptop.
I experimented with quite a few local LLMs too and granted, some perform a lot better than others, but they all have the same major issues. They don’t get smarter, they just produce the same nonsense faster (or rather often it feels like they are just more verbose about the same nonsense).
I don’t know what to tell you. I have them successfully compiling tables of search outputs to compare different things for method development and generating code, saving me hours of work each week. It all needs to be checked, but the comparison comes with links and the code is proofread and benchmarked. For most of what I do it’s really just a jacked up search engine, but it’s able to scan webpages faster than me and that saves a lot of time.
As a hobby, I also have it reading old documents that are almost illegible and transcribing them pretty well.
I really don’t know what you’re doing that you’re just getting nonsense. I’m not.
One other comment pointed me at one issue that might be a major difference. Is the code you generate in one of those ultra-verbose languages like Java where we had basically IDEs generating code from much shorter descriptions already 20 years ago? I could see LLMs doing well with those.
I tend to try to generate code mostly in Rust or sometimes shell or config files or DSL for various programs and 99% of the time the code does not even come close to what I wanted it to do, mainly because it just hallucinates itself some library interfaces that do not exist.
Not super common or super niche. I use R. And it completely made up code a year ago. Sometimes I still does, but less. And when I ask it for citations it can make shit up too. I really stand by the assertion that it needs a lot of babysitting.
But, between it getting better and me getting better at asking and some patience, I get what I want. But, it does require a lot of fine tuning and patience. But its still just faster than googling. And I could see the argument that the models haven’t improved but that they just have access to search engines now and that I’m mostly using them and a search engine. And sometimes they’re so whacked out I’ll ask them to search for something but theyll tell me they don’t have access to the internet and they’re so absolutely convinced of that that I have to close that chat and start a new one.
If you feed it in documentation or ask it to search for its answers in substack (or really just whatever search constraints you want) and then tell it to give you the links it used, you might have a better time. This forces it to look up an answer instead of hallucinate one. And when it gives me code, more complicated things usually fail pretty hard at first and I have to feed it the error output for a few rounds and guide it a lot.
brucethemoose@lemmy.world
on 29 Apr 16:36
collapse
It can be grounded in facts. It’s great at RAG. But even alone, Gemini 2.5 is kinda shockingly smart.
…But the bigger point is how Google presents it. It shouldn’t be the top result of every search just thrown into your face, it should be a opt-in, transparent, conditional feature with clear warnings, and only if it can source a set of whitelisted, reliable websites.
After just trying it again a few times today for a few practical problems that it not only misunderstood at first completely and then gave me a completely hallucinated answer to every single one I am sorry, but the only thing shocking about it is how stupid it is despite Google’s vast resources. Not that stupid/smart really apply to statistical analysis of language.
brucethemoose@lemmy.world
on 29 Apr 16:57
collapse
Gemini 2.5? Low temperature, like 0.2?
The one they use in search is awful, and not the same thing. Also, it’s not all knowing, you gotta treat it like it has no internet access (because generally it doesn’t).
I use it for document summarization and it works well. I use Paperless-ngx to manage documents, and have paperless-ai configured to instantly set the title and tags using Gemini as soon as a new document is added.
I chose Gemini over OpenAI since Google’s privacy policy is better. I’m using the paid version, and Google says data from paid users will never be used to train the model. Unfortunately I don’t have good enough hardware to run a local model.
Sad its results are less censored due to the others pairing with corporations.
suicidaleggroll@lemm.ee
on 29 Apr 16:36
nextcollapse
I abandoned Google when they started throwing shopping links at the top of every search, even when searching for things that have no relevance to shopping, and they started artificially promoting scams and paid material above actual results.
Google Search was best around 10-15 years ago when their only focus was providing the best results they could (remember when you could actually click the top result and you would be taken to the most applicable page instead of some unrelated ad or scam?). Now their focus is on providing the best product possible for their actual customers (paid advertisers) even when it means trashing their own product in the process.
And if tech people are no longer recommending it, or actively recommending against it, it is possible to get people to start switching. This is largely how chrome became so popular in the first place, and that required getting people to change from the default option.
I switched when the answers I got started to become bullshit. I’d google a simple question just to double check if it was correct, but it gave me something completely different. Something so out of the realm of possibility that I was baffled.
I check the sources for the answer and they were not even related. After that I started paying more attention to how messed up google had become, and I had enough.
Google scholar however is still something I need… even though I dislike American corporations.
turtlesareneat@discuss.online
on 29 Apr 18:54
nextcollapse
It’s kind of unreal that they took something that worked perfectly well for 25 years and then fucked it up entirely overnight, for no good reason.
and the nature of their 1000s of experiments going at the same time isn’t methodical at all. they don’t actually know what iteration of google search was the best and most useful one. there’s no going back to what worked because at no point did they ever know what worked. the modern shitty google search is the best version we will ever get to use again.
google embedded them as a core piece of information infrastructure and then demolished themselves. now our information networks fundamentally do not work anymore
I’m not going to defend google, but it’s important to understand the nature of the beast. It’s not possible to simply revert back to an older version of search ranking because it is a constant cat-and-mouse game between google developers and malicious actors trying to game the system with their shitty websites.
WanderingThoughts@europe.pub
on 29 Apr 19:55
collapse
They did that to drive up short term ad revenue and it worked, and joy was in their greedy little hearts. They also did figure out that poisoning search results drives away users, and that search is kinda the fastest gateway drug to their entire ecosystem.
But they’re stuck. Fixing search would lower their ad revenue, and stock holders would kill them for that.
cantstopthesignal@sh.itjust.works
on 29 Apr 22:01
collapse
Also, the way SEO is gamed, it’s really hard to unenshittify the internet. It’s not just their ads, it’s everyone making fake bullshit that pulls the right levers to get to the top of a search.
cantstopthesignal@sh.itjust.works
on 29 Apr 22:00
collapse
Why would I google something when the first answer is an inferior AI to chatgpt. I’ll just use a better AI.
Jaberw0cky@lemmy.world
on 29 Apr 19:17
nextcollapse
I stopped using Google at the same time I closed my accounts with Facebook, instagram, Reddit and Amazon. Currently I’m using Ecosia which I think is German. I’m dumping all the US companies I can based on all the Trump crap. It is taking time and effort but I should be able to actually close the Google account soon and I replaced windows with Linux on all but one of my PCs.
Yeah but don’t let perfect be the enemy of good. Among other things, they just recently announced that they’re starting to build an alternative index with Qwant.
Late to reply but: from what I heard, Kagi is really good at search (at least when it comes to filters and the such, not sure about base results), but it’s paid so that’s a disadvantage (and also a plus since you don’t get ads nor is your data being sold off), and it’s US based. DuckDuckGo also uses Bing but I guess they have more sources than just them. In my experience it’s about the same, maybe a bit better on DDG, both have shebangs (! for DDG, # for Ecosia) which I find extremely useful.
I’m using Qwant, used Ecosia before. Really OK for most stuff. I still revert to googling occasionally - mainly for local businesses on maps and sometimes shopping results. But I agree, don’t let perfect be the enemy of good, well said.
Madbrad200@sh.itjust.works
on 29 Apr 21:48
nextcollapse
In Ecosia’s settings you can set it to primarily use Google results instead of Bing. Makes it a lot more enjoyable to use.
PalmTreeIsBestTree@lemmy.world
on 30 Apr 03:19
collapse
Google maps is still the best for looking up local businesses and reviews. I wish people would go back to using a modern version of yellow pages.
Lost_My_Mind@lemmy.world
on 29 Apr 20:51
nextcollapse
I’ve never heard of Ecosia, but I don’t understand your logic on this.
Problem: Google bad!
Solution: Don’t use google, use Ecosia instead.
Error: Ecosia also uses google.
How is this a good move? If anything it’s just a lateral move with the same problem.
Late reply but: Ecosia lets you choose between Google or Bing, but I guess that isn’t really your point. The main thing is it anonymises your searches, so you do get some additional privacy. Additionally, the hope is that one day they build their own index. They just recently announced that they’re gonna start one with Qwant.
Venus_Ziegenfalle@feddit.org
on 30 Apr 07:09
collapse
You’re right but it’s still important to point out in this case since an American index would be subject to any American censorship law. It’s better to use Ecosia than Google for sure but we still gotta be aware of the type of bias we’re working with.
I hope that goes well for them. It’s hard and extremely expensive, which is why there’s so few good search engines and half of them just use Bing’s API.
Do you have a source for that? I think it’s nowhere near 95% of sites given there’s several major providers that aren’t AWS or Cloudflare (eg Hetzner, Google Cloud, DigitalOcean, Wordpress.com, and a bunch more)
I wouldn’t say dogshit, I would say the results are good enough for most purposes, not great, my priority just as the moment is dumping USA based companies and I thought Ecosia planting trees was a nice touch. Although as someone else pointed out it is still possibly using crawlers from google and bing in the background. I’m still keen to try out other search engines.
Definitely miss the good ol’ days where it was optimized to give the best results. Same goes for Netflix recommendations back in the the DVD mailer days…
stellargmite@lemmy.world
on 29 Apr 20:06
nextcollapse
I ditched it around 2014 when I noticed it had effectively become the yellow pages. Its pretending to be one thing to the ‘user’ when its actually serving someone else. This is transparent of course but the balance/compromise or tradeoff of it still providing some utility to the user despite this is what may vary for different people. My threshold was low. That and the privacy violations. Unfortunately its a corporation.
For a growing number of users, we can provide search results and ads from Google. For the eligible users, we will provide Google results by default. When your search results and ads are provided by Google, Google will use essential cookies and local storage to help defend against fraudulent traffic. Beyond this, the cookies Google uses will depend on where you are searching from: If you search from the EU, UK or certain US states (for example California), Google will not set additional cookies without your consent. If you search from elsewhere, Google may place additional cookies and the functionality of these cookies will depend on whether you have a Google account
Seems like using Ecosia is about as private as using Google directly.
Google’s been garbage for years now, I kind of miss Copernic Pro which is what I used before Google it searched all the search engines available and combined and resorted all the results.
Google was perfect at launch but in recent years it’s worse than Yahoo!.
threaded - newest
[the chart shows stats for American Google, American Bing, Russian Yandex, American Yahoo!, American DuckDuckGo, and Other]
Yeah, that statement wasn’t supported by the data at all. It seemed to only be included as a way to link to their other articles about European alternatives and de-Googling.
Exceptionally good at finding torrent sites and other piracy outlets, because they aren’t working hand-in-glove with American broadcasters to censor and shadowban these links. Google, Bing, DDG, and the other American mainline search sites all focus on feeding end-users into a discrete set of Web2 mega-site sponsors. Yandex uses the older web crawlers and indexing tools, so it gives more honest (abet fuzzier and less reliable) results. And since nobody really gives a shit about Yandex, the efforts to game its algorithm have been comparatively minimal.
Yandex also has the benefit of being relatively English-friendly, while other popular non-English search sites like Baidu, Qwant, and Naver don’t cater too quite so freely.
Thanks for mentioning Yandex, bringing is back onto my radar.
And they have really good products - the Navigator is great, and Yandex Music was better than Spotify (until the war started and a lot of labels/artists disappeared).
I’m not using their products now as I don’t want to feed the government, but they do(did?) some great stuff.
Who still uses search engines to find torrent, though?
Sites periodically get taken down or rendered less than useful. Especially for live streaming.
Yandex was invaluable when I was looking for Olympics streams, for instance. Also really depends on which communities are hosting to which torrent sites. I found nyaa.si off Yandex, because I couldn’t find the anime I was looking for on 1337x.to.
It’s been common ever since magnet links were created, since you can post a magnet link anywhere (even in a plain text file) rather than having to upload a .torrent file somewhere like in the old days.
What would be a good EU alternative?
Qwant or Ecosia.
Thanks!
.
the discussion is about search engines, not browsers.
I almost never use them now.
Google been degrading as time has gone on. The other search engines (like all of them) are getting or surpassing google in certain subjects. AI has really made them look like fools in all of this. Googles AI sucks for results and (while I dont like it) others are using chatgpt for search results.
I would say it is just the opposite. Google used to be good before they tried to post-process results in this extreme way and AI is just an even more extreme way to do that. ChatGPT and all the other LLMs just increase the noise to signal ratio (noise coming first because there is so much more of it than signal these days).
I actually agree. Others use chatgpt for just about everything. I dont like it for many…many reasons.
Google was much better pre-2020.
People use chat gpt to tell them what to buy. We are doomed. Our brains are about to shrink to the size of a pea in 10 years. All is going to plan for the elite.
Its funny because you can totally run good LLMs on local systems but people are just going to chat because its what they know and its easy to work with. Like I get it, but they are starting to put ads in the prompts now.
Google was best in the 2000s, but things were different back then. SEO spam wasn’t really a thing yet, there were far fewer websites, and most online discussions were archived and searchable (compared to today where there’s platforms like Discord that aren’t indexable in search engines at all).
You're even a verified user. So you must be right. 😝
Didn’t think Lemmy had such a concept.
It was just a joke, because of his avatar.
I work in an education setting and in the last month, Google started preloading the contents of other sites directly on the search page. It is wreaking havoc when combined with our blocking tools because kids will do a Google search for something innocuous and the page will immediately get blocked because it tried to load a result from Reddit or coursehero or something else we have blocked.
It’s incredibly frustrating.
If you have to stick with google, you can use udm=14. tedium.co/2024/…/google-web-search-make-default/ You can set it as default search provider.
Thanks for the tip
This is available in the UI too - there’s a tab labeled “Web”. Sometimes it’s hiding under “More”.
Adding it to the search provider URL is a good idea though.
Does that work in Firefox? Seems like I don’t have the ability to edit the address.
Yeah, even on mobile you can do it.
Amazing, thank you!
time to switch to Qwant, ecosia, or duckduckgo
Does that mean any search (AI insight notwithstanding) will get blocked if it includes a Reddit, Coursera or something on the blocklist result at all?
Because if yes, that’s much more than just asinine. It’s basically blocking entire search topics due to the sheer fact that Reddit will appear on the furst page of Google a lot.
That is exactly what is happening. They type in the search query on their Chromebook (for example, “why do dry erase markers float”, the results page flashes for a second and then the “this page is blocked” screen comes up saying they were blocked from Reddit, et al. Without them clicking on any search results.
And 2 kids can do the exact same search at the same time and get blocked for different sites or only one will get blocked.
Somewhat OT but why do you block coursera?
We have a different person that manages our block list so I don’t know the reason for all the blocks. That said, we block www.coursehero.com not www.coursera.org
Oh I thought it was a typo. I have no idea what coursehero is.
I will say coursera is awesome for their calculus classes. (Or they were like 10+ years ago anyway.) Which caught me up after a long hiatus from college when I returned to finish.
To be fair, I don’t know that we don’t block coursera as well but the block I specifically saw was for a course hero. And some of the blocks that we have on are for security reasons over reliability reasons. We have to be hyper cautious about the students leaking any potentially PHI and some of the Google sign in setups are less secure than others.
Google search has sucked for a couple years now. DuckDuckGo is better for everything except maps.
Thats because for some ungodly reason they use Apple Maps. Not sure why they dont integrat with an OpenStreetMaps like service. At least that way users can start contributing to fill the gaps
Mapy is also a great European alternative, based on OpenStreetMaps
That’s been my go to since they started. There is/was a challenge using them when we evaluated them a while back with forcing safe search reliability if I recall the reason.
I went long enough without using Google (probably a year-ish) that, when I accidentally made a Google search a few days ago, it was a jarring experience.
It felt wrong the same way other search engines did when I first deGoogled. It was kind of nice actually.
The irony is Gemini is really good (like significantly better than ChatGPT), and cheap for them (no GPUs needed), yet somehow they made it utterly unbearable in search.
“Significantly better than ChatGPT” and “Good” aren’t the same. Like ipecac is significantly better to drink than sewage water.
Gemini is really good at confidently talking nonsense but other than that I don’t really see where you get the idea that it is good. Mind you, that isn’t much better with the other LLMs.
So it’s really good at the thing LLMs are good at. Don’t judge a fish by it’s ability to climb a tree etc…
No, it is mediocre at best compared to other models but LLMs in general have a very minimal usefulness.
I get the desire to say this, but I find them extremely helpful in my line of work. Literally everything they say needs to be validated, but so does Wikipedia and we all know that Wikipedia is extremely useful. It’s just another tool. But its a very useful tool if you know how to apply it.
But Wikipedia is basically correct 99% of the time on basic facts if you look at non-controversial topics where nobody has an incentive to manipulate it. LLMs meanwhile are lucky if 20% of what they see even has any relationship to reality. Not just complex facts either, if an LLM got wrong how many hands a human being has I wouldn’t be surprised.
LLMs with access to the internet are usually about as factually correct as their search results. If it searches someone’s blog, you’re right, the results will suck. But if you tell it to use higher quality resources, it returns better information. They’re good if you know how to use them. And they aren’t good enough to be replacing as many jobs as all these companies are hoping. LLMs are just going to speed up productivity. They need babysitting and validating. But they’re still an extremely useful tool that’s only going to get better and LLMs are here to stay.
That is the thing, they are not “only going to get better” because the training has hit a wall and the compute used will have to be reduced since they are losing money with every request currently.
Technology these days works in that they always lose money at the start. Its a really stupid feature of modern startups IMO. Get people dependent and they make money later. I don’t agree with it. I don’t really think oir entire economic system is viable though and that’s another conversation.
But LLMs have been improving exponentially. I was on board with everything you’re saying just a year ago about how they suck and they’re going to hit a wall even. But the don’t need more training data or the processing power. They have those and now they’re refining the LLMs. I have a local LLM on my computer that performs better than chat GPT did a year ago and it’s only a few GB. I run it on a shitty laptop.
I experimented with quite a few local LLMs too and granted, some perform a lot better than others, but they all have the same major issues. They don’t get smarter, they just produce the same nonsense faster (or rather often it feels like they are just more verbose about the same nonsense).
I don’t know what to tell you. I have them successfully compiling tables of search outputs to compare different things for method development and generating code, saving me hours of work each week. It all needs to be checked, but the comparison comes with links and the code is proofread and benchmarked. For most of what I do it’s really just a jacked up search engine, but it’s able to scan webpages faster than me and that saves a lot of time.
As a hobby, I also have it reading old documents that are almost illegible and transcribing them pretty well.
I really don’t know what you’re doing that you’re just getting nonsense. I’m not.
One other comment pointed me at one issue that might be a major difference. Is the code you generate in one of those ultra-verbose languages like Java where we had basically IDEs generating code from much shorter descriptions already 20 years ago? I could see LLMs doing well with those.
I tend to try to generate code mostly in Rust or sometimes shell or config files or DSL for various programs and 99% of the time the code does not even come close to what I wanted it to do, mainly because it just hallucinates itself some library interfaces that do not exist.
Not super common or super niche. I use R. And it completely made up code a year ago. Sometimes I still does, but less. And when I ask it for citations it can make shit up too. I really stand by the assertion that it needs a lot of babysitting.
But, between it getting better and me getting better at asking and some patience, I get what I want. But, it does require a lot of fine tuning and patience. But its still just faster than googling. And I could see the argument that the models haven’t improved but that they just have access to search engines now and that I’m mostly using them and a search engine. And sometimes they’re so whacked out I’ll ask them to search for something but theyll tell me they don’t have access to the internet and they’re so absolutely convinced of that that I have to close that chat and start a new one.
If you feed it in documentation or ask it to search for its answers in substack (or really just whatever search constraints you want) and then tell it to give you the links it used, you might have a better time. This forces it to look up an answer instead of hallucinate one. And when it gives me code, more complicated things usually fail pretty hard at first and I have to feed it the error output for a few rounds and guide it a lot.
It can be grounded in facts. It’s great at RAG. But even alone, Gemini 2.5 is kinda shockingly smart.
…But the bigger point is how Google presents it. It shouldn’t be the top result of every search just thrown into your face, it should be a opt-in, transparent, conditional feature with clear warnings, and only if it can source a set of whitelisted, reliable websites.
After just trying it again a few times today for a few practical problems that it not only misunderstood at first completely and then gave me a completely hallucinated answer to every single one I am sorry, but the only thing shocking about it is how stupid it is despite Google’s vast resources. Not that stupid/smart really apply to statistical analysis of language.
Gemini 2.5? Low temperature, like 0.2?
The one they use in search is awful, and not the same thing. Also, it’s not all knowing, you gotta treat it like it has no internet access (because generally it doesn’t).
The one they use on gemini.google.com (which is 2.5 right now but was awful in earlier versions too).
Try it here instead, set the temperature to like 0.1 or 0.2, and be sure to set 2.5 Pro:
aistudio.google.com
It is indeed still awful for many things. It’s a text prediction tool, not a magic box, even though everyone advertises it kinda like the later.
I use it for document summarization and it works well. I use Paperless-ngx to manage documents, and have paperless-ai configured to instantly set the title and tags using Gemini as soon as a new document is added.
I chose Gemini over OpenAI since Google’s privacy policy is better. I’m using the paid version, and Google says data from paid users will never be used to train the model. Unfortunately I don’t have good enough hardware to run a local model.
I had that happen too. Couldn’t find something with DDG. Hopped over to Google and was shocked at how completely unusable it was.
I’ve fully switched to ecosia. I much prefer their efforts, and they seem to fund decent projects unlike a lot of other carbon offset companies.
Yandex is better than any American service by far imo.
…
Sad its results are less censored due to the others pairing with corporations.
I abandoned Google when they started throwing shopping links at the top of every search, even when searching for things that have no relevance to shopping, and they started artificially promoting scams and paid material above actual results.
Google Search was best around 10-15 years ago when their only focus was providing the best results they could (remember when you could actually click the top result and you would be taken to the most applicable page instead of some unrelated ad or scam?). Now their focus is on providing the best product possible for their actual customers (paid advertisers) even when it means trashing their own product in the process.
they also ruined their own platform by creating and encouraging an entire business around gaming search results.
15+ years ago you could search for an error code, or an error message, or a part number and actually find it.
Everybody is using ai of course.
Ditched Google for DuckDuckGo in like 2019 and used that for a long time until Kagi came out and now I use that and I’m quite happy with it.
Didn’t read the article but use Qwant since Trump.
Or Ecosia
Who knew if your product was mostly shit people would stop using it?
.
77% still using it though
Sure, but this is a process that takes time; that said, he trend is downward and that will likely continue unless things improve.
The recent duckduckgo ad campaign will surely help rescue googlers and so does my mission to ensure everyone I know doesn’t use google search.
.
And if tech people are no longer recommending it, or actively recommending against it, it is possible to get people to start switching. This is largely how chrome became so popular in the first place, and that required getting people to change from the default option.
This might be very non-linear, like ice breaking under your feet.
I’m glad to contribute to that metric.
Nice to see this choice is making an actual difference.
ohthankgod!
I switched when the answers I got started to become bullshit. I’d google a simple question just to double check if it was correct, but it gave me something completely different. Something so out of the realm of possibility that I was baffled.
I check the sources for the answer and they were not even related. After that I started paying more attention to how messed up google had become, and I had enough.
Google scholar however is still something I need… even though I dislike American corporations.
It’s kind of unreal that they took something that worked perfectly well for 25 years and then fucked it up entirely overnight, for no good reason.
Stick in bike spokes meme.
and the nature of their 1000s of experiments going at the same time isn’t methodical at all. they don’t actually know what iteration of google search was the best and most useful one. there’s no going back to what worked because at no point did they ever know what worked. the modern shitty google search is the best version we will ever get to use again.
google embedded them as a core piece of information infrastructure and then demolished themselves. now our information networks fundamentally do not work anymore
I’m not going to defend google, but it’s important to understand the nature of the beast. It’s not possible to simply revert back to an older version of search ranking because it is a constant cat-and-mouse game between google developers and malicious actors trying to game the system with their shitty websites.
Enshitification.
They did that to drive up short term ad revenue and it worked, and joy was in their greedy little hearts. They also did figure out that poisoning search results drives away users, and that search is kinda the fastest gateway drug to their entire ecosystem.
But they’re stuck. Fixing search would lower their ad revenue, and stock holders would kill them for that.
Also, the way SEO is gamed, it’s really hard to unenshittify the internet. It’s not just their ads, it’s everyone making fake bullshit that pulls the right levers to get to the top of a search.
Why would I google something when the first answer is an inferior AI to chatgpt. I’ll just use a better AI.
I stopped using Google at the same time I closed my accounts with Facebook, instagram, Reddit and Amazon. Currently I’m using Ecosia which I think is German. I’m dumping all the US companies I can based on all the Trump crap. It is taking time and effort but I should be able to actually close the Google account soon and I replaced windows with Linux on all but one of my PCs.
Ecosia still uses American services though - they use Google, Bing, Yahoo and Wikipedia for search results.
Yeah but don’t let perfect be the enemy of good. Among other things, they just recently announced that they’re starting to build an alternative index with Qwant.
Definitely true. I’ll have to try it out. Is Ecosia better than DuckDuckGo or Kagi?
Late to reply but: from what I heard, Kagi is really good at search (at least when it comes to filters and the such, not sure about base results), but it’s paid so that’s a disadvantage (and also a plus since you don’t get ads nor is your data being sold off), and it’s US based. DuckDuckGo also uses Bing but I guess they have more sources than just them. In my experience it’s about the same, maybe a bit better on DDG, both have shebangs (! for DDG, # for Ecosia) which I find extremely useful.
I’m using Qwant, used Ecosia before. Really OK for most stuff. I still revert to googling occasionally - mainly for local businesses on maps and sometimes shopping results. But I agree, don’t let perfect be the enemy of good, well said.
In Ecosia’s settings you can set it to primarily use Google results instead of Bing. Makes it a lot more enjoyable to use.
Google maps is still the best for looking up local businesses and reviews. I wish people would go back to using a modern version of yellow pages.
I’ve never heard of Ecosia, but I don’t understand your logic on this.
Problem: Google bad!
Solution: Don’t use google, use Ecosia instead.
Error: Ecosia also uses google.
How is this a good move? If anything it’s just a lateral move with the same problem.
Eventually they’ll stop using G and MAYBE they have better impact in the climate. Why be a fucking prick about it?
.
Late reply but: Ecosia lets you choose between Google or Bing, but I guess that isn’t really your point. The main thing is it anonymises your searches, so you do get some additional privacy. Additionally, the hope is that one day they build their own index. They just recently announced that they’re gonna start one with Qwant.
You’re right but it’s still important to point out in this case since an American index would be subject to any American censorship law. It’s better to use Ecosia than Google for sure but we still gotta be aware of the type of bias we’re working with.
What about DuckDuckGo?
.
At this point Bing is better than Google, so using a privacy based skin of Bing seems like a good idea.
ecosia is developing their own engine apparently
I hope that goes well for them. It’s hard and extremely expensive, which is why there’s so few good search engines and half of them just use Bing’s API.
95+% of all websites you visit are hosted on AWS or use Cloudflare.
But that’s their decision, not yours.
Do you have a source for that? I think it’s nowhere near 95% of sites given there’s several major providers that aren’t AWS or Cloudflare (eg Hetzner, Google Cloud, DigitalOcean, Wordpress.com, and a bunch more)
Same here! Loved the switch to Linux.
.
I wouldn’t say dogshit, I would say the results are good enough for most purposes, not great, my priority just as the moment is dumping USA based companies and I thought Ecosia planting trees was a nice touch. Although as someone else pointed out it is still possibly using crawlers from google and bing in the background. I’m still keen to try out other search engines.
.
interesting. looks like yandex and yahoo are the ones who took it.
Makes sense when you consider that Yandex is owned by a Russian oligarch, and Google has been fined more money than exists in Russia.
I still include Google results in my Searx.
Definitely miss the good ol’ days where it was optimized to give the best results. Same goes for Netflix recommendations back in the the DVD mailer days…
I ditched it around 2014 when I noticed it had effectively become the yellow pages. Its pretending to be one thing to the ‘user’ when its actually serving someone else. This is transparent of course but the balance/compromise or tradeoff of it still providing some utility to the user despite this is what may vary for different people. My threshold was low. That and the privacy violations. Unfortunately its a corporation.
Consider Kagi far superior.
I ditched Google and Gmail.
Same trash.
.
Privacy company offers privacy focused LLMs for users that would’ve otherwise paid for ChatGPT. Unbelievable
Did you actually ever try Kagi or do you just want to spit uninformed delusions?
Those LLMs can never be privacy focused when they are still trained in stolen data, stolen art, stolen research and stolen models.
Every. Single. One.
You are confusing morality and copyright with user privacy.
You also failed to access if you’ve actually tried the search engine that you are no quick to spit nonsense about.
Ok so what’s your alternative then?
.
I ended up having to install a CalDAV client on my other people’s Android phone.
<img alt="I’m Doing My Part!" src="https://media1.tenor.com/m/zJvexdmTjA4AAAAC/im-doing-my-part-serious.gif">
!lemmysilver
Thank you for voting. You can vote again in 24 hours. leaderboard
Been using Ecosia for up to a year now probably. I like it and I don’t need anything else.
Is Ecosia any good for the privacy-minded folks among us?
It’s more privacy focussed than Google, but less than others. Personally I use Ecosia at work and Qwant at home.
https://www.ecosia.org/privacy
Seems like using Ecosia is about as private as using Google directly.
Google’s been garbage for years now, I kind of miss Copernic Pro which is what I used before Google it searched all the search engines available and combined and resorted all the results.
Google was perfect at launch but in recent years it’s worse than Yahoo!.
If you’re tech savvy, look into selfhosting SearXNG.
I think there are public instances as well.
Would it run on a Raspberry pi 5?
I’m running one on pi5 with no issue. It takes less than 5 minutes to install one under docker.
No effing way! I got a 16gb model, doing this right………now.
Yeah it doesn’t use many resources.