|
I basically use startpage everywhere but no search engine is perfect anyway. It is good google is feeling the heat already in some fashion so they don't just focus on monopolistic behavior all day. However - google in the sentence above can be swapped with a majority of companies, let alone tech companies in specific.
|
#
?
Feb 12, 2023 14:46
|
|
|
|
| # ? Jun 3, 2024 17:32 |
|
|
Google Search Is Dying
|
|
|
#
?
Feb 12, 2023 14:51
|
|
|
The single biggest problem I have with google for tech queries is the loving "did you mean __" and "including results for __". Everything has to go in quotes now, and even then it still thinks it knows better than me. It's a big problem whenever you are searching for anything less common. I've sometimes had good luck with Yandex for tech queries, FWIW.
|
|
#
?
Feb 12, 2023 15:00
|
|
|
feedmyleg posted:This is pretty universally recognized from what I've seen. Results are hot garbo and I'm really excited for them to launch a ton of new search products that do not in any way address the core issues. (theatrically putting my hand to my ear) I'm hearing that our customers want a 'search experience', where they can type incredibly simple what-is-the-capital-of-France / how-tall-is-the-Empire-State-Building / what-is-the-current-time-in-Tokyo questions to an AI bot, which we've now decided is our core business - following up on our successes in the gaming, VR, and social media 'spaces'
|
|
#
?
Feb 12, 2023 15:06
|
|
so I could say that I finally figured out what this god damned cube is doing. Get well Lowtax.
|
I saw someone mention kagi.com, maybe in this thread. I've been using it for about a month now and it's good enough to replace Google search for me. The extra bells and whistles of being able to block sites from results as well as prioritize results from particular sites is really nice.
|
|
#
?
Feb 12, 2023 15:37
|
|
|
withoutclass posted:I saw someone mention kagi.com, maybe in this thread. I've been using it for about a month now and it's good enough to replace Google search for me. The extra bells and whistles of being able to block sites from results as well as prioritize results from particular sites is really nice. Same, I've been using kagi for about a week now with no problems.
|
|
#
?
Feb 12, 2023 15:45
|
|
|
sgbyou posted:Same, I've been using kagi for about a week now with no problems.
|
|
#
?
Feb 12, 2023 15:51
|
|
|
Boris Galerkin posted:Although it could also go wrong and give wrong information, as with Google’s recent JWST fiasco. The fundamental issue for me with this is: 1. Yes, snippets of relevant info at a glance are useful. 2. In industry, we are generally pretty good at full text document search and extracting relevant info from something as well-structured as Wikipedia. This type of feature is reliable and suits the needs for most use cases. 3. LLMs are entirely non deterministic and probabilistic, which means they may accurately summarize the information that you want. But they also might probabilistically choose to summarize entirely wrong information. Regardless, it's going to look coherent enough that it's hard to tell. I work in search. I feel like I understand the tech. I do not see how this moves the expectation of how search operates for the average consumer in the way that the hype suggests. The way that Google and Bing are betting the farm feels like it's all about investors, not about addressing a real need. I have not spoken to a single person who isn't mainlining tech news who gives a poo poo, or even is aware, about any of this.
|
|
#
?
Feb 12, 2023 16:03
|
|
|
cat botherer posted:The single biggest problem I have with google for tech queries is the loving "did you mean __" and "including results for __". Everything has to go in quotes now, and even then it still thinks it knows better than me. It's a big problem whenever you are searching for anything less common. The ones that drive me nuts are where if you search something that's a degree of separation from something more common, it just gives you the more common thing. I'm endlessly searching for info about step two of a process only to get results about step one, or articles about how the process is possible. This is most infuriating when the issue is that some menu item in loving Android straight-up isn't there. I Google "the fooble the narble option is missing" and it acts like I googled "fooble the narble" and get endless pages showing pictures of a menu option I don't have.
|
|
#
?
Feb 12, 2023 16:07
|
|
|
cat botherer posted:Charging for search is a pretty big ask for most people though. Of course that does better enable them to not be lovely. It's an interesting model. Neeva is another one trying it out. I get the intent, but it seems niche, which in turn makes it hard to get enough traction to make results amazingly good. At any rate, all of these ideas of shaking up search seem to not address some of the more fundamental issues - like how a lot of people just don't use search engines, and instead use Facebook/TikTok/Reddit to find what they want.
|
|
#
?
Feb 12, 2023 16:14
|
|
|
Blue Footed Booby posted:The ones that drive me nuts are where if you search something that's a degree of separation from something more common, it just gives you the more common thing. I'm endlessly searching for info about step two of a process only to get results about step one, or articles about how the process is possible. yeah, this is infuriating. im pretty sure this got a lot worse when search had Bidirectional Encoder Representations from Transformers (BERT) added to it in 2019. since then (probably before that too) search queries are going through at least one level of interpretration which tends to homogenize queries into some common streams. here's one of their blogs from 2019 on adding BERT to search. the T in BERT is the same as the T in ChatGPT. natural language processing has been making search worse for niche purposes for quite while https://blog.google/products/search/search-language-understanding-bert/
|
|
#
?
Feb 12, 2023 16:16
|
|
|
Walh Hara posted:Speaking about search engines: This week I had two situations where google failed to point me to an answer of a fairly technical, but not super complex, question. It really gave me the impression that the search engine become worse somehow. A follow up. I just tried this in ChatGPT. I now understand why google is scared. First question:  Second question:  The answer to the second question is especially impressive (although not 100% perfect).
|
|
#
?
Feb 12, 2023 17:29
|
|
|
Walh Hara posted:A follow up. I just tried this in ChatGPT. I now understand why google is scared. You probably could have just searched StackOverflow directly and gotten those same results, since that's where ChatGPT is pulling them from anyway - but with the extra advantage that you'd been able to see the votes and comments so you'd know if everyone's saying "this solution is poo poo that doesn't work" or "this solution is for an old incompatible version from eight years ago".
|
|
#
?
Feb 12, 2023 17:38
|
|
|
Saw a video the other day - guy who's been doing Mario 64 hacking has done all sorts of engine improvements and optimizations thanks to the source code leaks basically tried to get ChatGPT to optimize some of his code: https://www.youtube.com/watch?v=9QZlQMpNk-M It's really neat, but one of the takeaways seems to be that ChatGPT can be -confidentially wrong- and sometimes just suggest things that aren't possible or maybe don't make sense. I don't think any of the optimizations really ended up helping Kaze, but it definitely seemed to produce results he seemed to think was worth trying, and the ability for him to go "okay how do I implement that idea" seemed interesting.
|
|
#
?
Feb 12, 2023 17:50
|
|
|
Main Paineframe posted:You probably could have just searched StackOverflow directly and gotten those same results, since that's where ChatGPT is pulling them from anyway - but with the extra advantage that you'd been able to see the votes and comments so you'd know if everyone's saying "this solution is poo poo that doesn't work" or "this solution is for an old incompatible version from eight years ago". I tried a few ways to formulate my question, but the stack overflow search failed to find a relevant answer for my second question. It's indeed very likely that the answer I needed is also on stack overflow somewhere, but I don't find it within 2 minutes searching. For the first question the answer is was looking for was an airflow reference documentation page ("it's the parameter 'retry'"), it's not a complicated problem that requires a stack overflow discussion.
|
|
#
?
Feb 12, 2023 19:16
|
|
|
feedmyleg posted:This is pretty universally recognized from what I've seen. Results are hot garbo and I'm really excited for them to launch a ton of new search products that do not in any way address the core issues. Nah Google's fine, just fine... *scrolls past doesn't of SEO-farming content mill sites that just rehash the same article and link to one another without answering my question*
|
|
#
?
Feb 12, 2023 19:48
|
|
|
Oxyclean posted:I don't think any of the optimizations really ended up helping Kaze, but it definitely seemed to produce results he seemed to think was worth trying, and the ability for him to go "okay how do I implement that idea" seemed interesting. I knew there was still room to improve on the plastic rubber duck. Instead of using fossil fuels once during production, you can burn them continuously to keep it going
|
|
#
?
Feb 12, 2023 19:51
|
|
|
Evil Fluffy posted:Nah Google's fine, just fine... What's especially frustrating is that you used to be able to block sites from appearing in results, but not anymore. There are very specific SEO spam sites that keep coming up, and I'd rather play whack-a-mole than just give up altogether
|
|
#
?
Feb 12, 2023 20:03
|
|
|
Don't forget that when you look for "product that does this", you're almost certain to be given a bunch of lists to paid solutions, with at most free limited plans - if you instead try finding a list on Wikipedia it will be more expansive and likely to include open source ones.
|
|
#
?
Feb 12, 2023 20:04
|
|
|
Ruffian Price posted:I knew there was still room to improve on the plastic rubber duck. Instead of using fossil fuels once during production, you can burn them continuously to keep it going Is AI that resource/energy intensive? I know it takes a good deal of processing power, but I have the impression it's very far from "crypto farm" levels.
|
|
#
?
Feb 12, 2023 20:11
|
|
|
My favorite part of every google search now is the first dozen hits being “best [search terms] of 2023” articles written by AIs trained on Amazon reviews.
|
|
#
?
Feb 12, 2023 20:31
|
|
|
Oxyclean posted:Is AI that resource/energy intensive? I know it takes a good deal of processing power, but I have the impression it's very far from "crypto farm" levels. Training the models are usually the worst parts ime. I haven't specifically worked with some of the new LLMs, but the vast majority of ANN stuff requires GPU computing to train models in less than a century's worth of CPU time. Assuming model training is an ongoing process, it's a lot of ongoing GPU resources, but it's not decentralized the same way that crypto mining was.
|
|
#
?
Feb 12, 2023 20:46
|
|
|
withak posted:My favorite part of every google search now is the first dozen hits being “best [search terms] of 2023” articles written by AIs trained on Amazon reviews. The one nice thing about these is that they'll give you enough brand names that you can feed back into a search, and it'll usually pull up a Wikipedia article showing you what your choices are, or to introduce into a reddit / [insert appropriate community here] search in pairs or trios to find comparisons.
|
|
#
?
Feb 12, 2023 20:49
|
|
|
Kestral posted:or to introduce into a reddit / [insert appropriate community here] search in pairs or trios to find comparisons. That's pretty funny because I saw that in this article posted earlier: People are tacking on reddit to their searches so they have a greater chance of getting something generated by a human.
|
|
#
?
Feb 12, 2023 22:42
|
|
|
Oxyclean posted:Is AI that resource/energy intensive? I know it takes a good deal of processing power, but I have the impression it's very far from "crypto farm" levels.  but compared to an index search, I would indeed call that a colossal waste. Especially if you've seen that giant loving initial prompt Bing's implementation uses for every question, no matter how simple
|
|
#
?
Feb 13, 2023 00:09
|
|
|
Oxyclean posted:It's really neat, but one of the takeaways seems to be that ChatGPT can be -confidentially wrong- and sometimes just suggest things that aren't possible or maybe don't make sense. https://youtu.be/bEghu90QJH4
|
|
#
?
Feb 13, 2023 00:20
|
|
|
Oxyclean posted:Is AI that resource/energy intensive? I know it takes a good deal of processing power, but I have the impression it's very far from "crypto farm" levels.
|
|
#
?
Feb 13, 2023 01:11
|
|
|
Foxfire_ posted:Think roughly a second or so of playing a video game on fancy graphics settings per inference Prediction goes forward through the model once, whereas training goes forward and backward dozens of times per example.
|
|
#
?
Feb 13, 2023 01:26
|
|
|
Rocko Bonaparte posted:That's pretty funny because I saw that in this article posted earlier: I do that for the majority of my searches now, or if not Reddit, then some appropriate community. The initial search might start as a general googling, but it almost inevitably gets to using site:reddit.com or an equivalent. It just works better, period.
|
|
#
?
Feb 13, 2023 02:12
|
|
|
cat botherer posted:Prediction isn’t so bad, but training can take as much as multiple weeks with the gpu going full tilt. I wonder what the difference in compiling is from initial to additional case addition. I wonder what difference that creates based on the weight.
|
|
#
?
Feb 13, 2023 02:49
|
|
|
Main Paineframe posted:You probably could have just searched StackOverflow directly and gotten those same results, since that's where ChatGPT is pulling them from anyway - but with the extra advantage that you'd been able to see the votes and comments so you'd know if everyone's saying "this solution is poo poo that doesn't work" or "this solution is for an old incompatible version from eight years ago". That's why I'm sticking to doing my own searches for a good long time. The context of the answer, and its discussion, is often at least as useful as the lines of code or whatever.
|
|
#
?
Feb 13, 2023 05:44
|
|
|
Rocko Bonaparte posted:People are tacking on reddit to their searches so they have a greater chance of getting something generated by a human. I also tack on "forums", still a lot of forums out there and outside tech related stuff they tend to have a lot of good info IMO. More than once I found the info I needed from some old forum thread where some dude I never knew had a similar problem.
|
|
#
?
Feb 13, 2023 08:20
|
|
|
His Divine Shadow posted:I also tack on "forums", still a lot of forums out there and outside tech related stuff they tend to have a lot of good info IMO. One thing I hate is a lot of hobbyist hardware/software projects sometimes live almost exclusively on niche discord servers. Especially game homebrew stuffs.
|
|
#
?
Feb 13, 2023 10:24
|
|
|
Ah Discord. I tried to register there, hit wrong key by accident and submitted without a password (can't believe it allowed it). Tried to reset my password so it wouldn't be nothing, was immediately flagged as suspicious behaviour by some AI. Account / phone number banned for life, no appeal it said.
|
|
#
?
Feb 13, 2023 10:51
|
|
|
BlueBlazer posted:I wonder what the difference in compiling is from initial to additional case addition. I wonder what difference that creates based on the weight. There is a bit of a wrinkle here with biases caused by the new data (possibly necessitating passing over old data more) but that's kind of separate issue. Sometimes it might be a good thing, for example if you are taking a pre-trained model and tuning (some) weights toward some more specialized set of examples.
|
|
#
?
Feb 13, 2023 15:30
|
|
|
Turns out Bings demo of its GPT search also had a load of mistakes, but unlike google no one noticed straight away. https://dkb.blog/p/bing-ai-cant-be-trusted It recommending a gay bar on the Mexico visit but not mentioning it was a gay bar gave me a good chuckle though. Mega Comrade fucked around with this message at 20:21 on Feb 13, 2023 |
|
#
?
Feb 13, 2023 20:19
|
|
|
Speaking of Bing, I saw this screenshot posted on HN which the OP says he asked Bing before yesterday's game. Which OK is hilariously funny.
|
|
#
?
Feb 13, 2023 20:42
|
|
|
His Divine Shadow posted:Ah Discord. I tried to register there, hit wrong key by accident and submitted without a password (can't believe it allowed it). So make another account?
|
|
#
?
Feb 13, 2023 21:00
|
|
|
MixMasterMalaria posted:So make another account? Not the OP but I tried making a second account sometime last year and IIRC it required a phone number. Like the account wouldn’t activate and/or let me join any servers without activating through a phone number. My memory could be wrong but the result was I couldn’t make a second working account without jumping through hoops I wasn’t willing to bother with.
|
|
#
?
Feb 13, 2023 21:42
|
|
|
|
| # ? Jun 3, 2024 17:32 |
|
|
Boris Galerkin posted:Speaking of Bing, I saw this screenshot posted on HN which the OP says he asked Bing before yesterday's game. I wonder if it grabbed the madden sim results or something.
|
|
#
?
Feb 13, 2023 21:55
|
|