Technically adept people would search using ordered keywords and phrases, which would have excellent results with Google 2006:
Dario Turkey NBA "never coming over"
But regular people agonized over this because to them a search was
What were some of the memes about Dario staying in turkey
Google put immense effort into calibrating search for regular people instead of engineers.
I've always felt that "proper" and responsible LLM use for search [0] would be two boxes: You can describe what you want in the first, and it'll propose search terms in the second, and then those get executed normally.
Yes, the "average user" [1] might not usually care about the second box... Until they need to because the query/results are wrong.
Showing them in tandem means:
1. Users are at least capable of learning through exposure.
2. Users may realize a key term can be added which the model could never have guessed.
3. Users may recognize a term in there that doesn't make sense, allowing them to detect a translation error.
4. If good search-terms leads to a bad outcome, it is possible for someone to report and diagnose it, rather than a fully black-box mystery.
_____
[0] Not just for websites, but also things like internal business software, or SQL queries.
[1] The average that might not exist. ( https://www.thestar.com/news/insight/when-u-s-air-force-disc... ) There are some features that everybody needs, just at different times.
And regarding “average doesn’t exist” - that’s true. But no company does a/b testing to land on average. I’d assume a good 85%-kinda pass rate for these type of experiments.
95% of users might not want the double-textbox, but 5% might, and some might want to use it someday.
I think, make enough of these decisions, and there's bound to be mounting friction for users with different preferences navigating your app.
Ideally, the app should offer a way for the user to intuitively configure the app to their own liking.
Ideally, yeah maybe, but why bother with extra implementation, support, costs and etc., when people fold and use the new way anyways?
I think the two-tiered input-output approach you propose makes sense. Allowing users to inspect and mutate lower-level languages allows the user to make modifications as needed. I think it's very much in the spirit of free software.
But then again, for text-based search specifically (search engines, notes, etc.), I think there is some value in querying the LLM directly, as it is able to fuzzy search by inspecting its weights. This results in a lesser degree of specificity, which allows for more false positives, but can maybe capture similar words (i.e., synonyms / typos / tenses) or higher-level semantic concepts.
Maybe they're just two different search algorithms, and the user should be able to choose between them.
Thanks for linking the article, I found it very interesting.
For a certain cohort (eg those currently between 35 and 45) who did any sort of grade school computer/library classes in the 90's/early2000's, 'chunking' or keyword searching was one of the main skills taught.
If you are properly lazy, you must expect or at least attempt to get results for the shorter query. Especially on a phone.
Many people who learned to search the Internet in early 2000s learned "boolean search queries" (terms connected with AND or OR). They're talking about the difference between boolean search queries and natural language search queries.
The kids who stopped getting that are probabaly old enough for a bachelors degree now. Today, the schools mostly just hand them an iPad and a Google Workspace sign-in and pretend everything is fine.
All those studies about GUI usability were done in the early 2000s. Before phones. When most people were just learning how to use a computer. The idea of options certainly was confusing, because "clicking on things" was even confusing.
Yet today, we have a populous which, more and more, grew up with computers. And loves to personalize. And wants options. So just keep that in mind, all you frontenders, config options aren't necessarily bad. Ancient studies have little relevance 24 years later.
Sorry to hijack a bit, but felt it was relevant to courses, teaching, and the state of time in 2000 vs now.
https://ometer.com/preferences.html
I'm not sure about you, but I find Gnome very easy to use, although I miss some features. Those features can be added with Gnome extensions that are maintained by third-parties. Meanwhile I find GIMP extremely hard to use, as it packs in too many features to allow their designers to figure out a good UX which doesn't break some component.
I wouldn't have taken keyboarding if I hadn't been forced to at the age of 8. I hated keyboarding classes. I am so glad I took them now. I hated doing boring research projects and looking crap up on Yahoo's Directory. I love that I have a basic understanding of how to rate and investigate a website's trustworthiness. Mfers don't even look at the "About" section anymore.
You were lucky to have that part. Mine, as I remember, was just Microsoft Office throughout.
Answering with AI slop is a choice.
The PageRank algorithm, and importantly, any algorithm which depends on page content and pages linking to each other and referencing each other was DoA, because it only worked in an adversary free context! The very second there was a real incentive to be at the top of Google's search results, their algorithms were going to be attacked and manipulated.
That makes "search" a neverending arms race. The very thing that launched Google as some amazing thing was doomed from the start. They've been laying hacks on top of hacks on top of hacks ever since. It's been shit basically since SEO has existed, which is what the "adversary" is.
Another good point for when google "died" was when they bought doubleclick. They were an ad company from that point on. There was never a good outcome possible after that. If ads were something that benefited you the user, you would have to pay for them.
If you haven’t read it from start to finish you should. If you have, what about it is insufficient?
Instead, the AI answered that it is "actually a famous example from internet culture used to describe how people use search engines. The phrase stems from discussions on tech forums like Hacker News regarding the differences between "keyword searching" and "natural language searching"".
Top of the search result page: "Did you mean: vim"
Felt like I was being trolled.
These days, we default to
"Day 3,449: google's still broken. what was i thinking?"
Google can do it, but they must appease the share holders.
Google works like an automatic black-box which tries to deduce your intention from 4-5 words you typed in hastily.
I believe any ordinary user is still pretty happy about Google search in its current iteration. My mom doesn't get a YouTube premium account because she says she likes the ads, for example.
So, from our bubble, Google might be regressing, but the picture might be very different for the majority of Google users.
Duckduckgo and brave search are definitely not. It's always g! after a wildly fuzzy first pass.
I use duckduckgo, mostly by habit at this point, and if i fail to find anything i go to claude and ask it.
Im getting closer and closer to just defaulting to the AI model search every day. Especially when I want to find something that I can list off 5+ semi accurate details about but cant remember the name of.
Maybe?
https://en.wikipedia.org/wiki/Post-it_note
>In 1968, Spencer Silver, a scientist at 3M in the United States, attempted to develop a super-strong adhesive. [...]
>Post-its were launched across the United States in 1980.[20][21] The following year, they were launched in Canada and Europe.[22] Post-it Notes as we know them were patented by Fry in 1993 as a "repositionable pressure-sensitive adhesive sheet material".[23]
https://en.wikipedia.org/wiki/Cyanoacrylate
The first thorough research went into them when looking for new clear plastics, and this route of making them was ruled out because it would rather stick to everything than ease the production of objects with nice optical properties. The story goes, it was so annoying to work with that it was initially shelved, and only years after being considered again and ruled out again for a different project, the utility of its reliable and fast bonding was fully appreciated.
"No true normie would be dissatisfied with this unhelpful response, really it's exactly what a true normie wants"
It reminds me of the motion controls we now have in video game controllers. In the 80s I remember my parents moving the controller as if that was going to help their Teris block or Mario go a little further than the D-pad alone. We laughed when people did this. Now, the controllers respond to what people had been doing for decades with no effect.
I suppose this is what technology should do, adapt to how people naturally use the thing, rather than trying to rigidly adhere to conventions users are expected to learn… that were only conventions due to the limits of technology when it was first developed.
I want to know who's behind some presentation of their "the thing" as fact.
I want to know who's to benefit from their presenting "the thing" as fact.
AI or not, my needs are still the same and so have to go right to the sources and judge for myself.
I think about that lots of times when I google these days. That he wasn't wrong, was just asking it 15 years too early.
And the impact on discourse has been, frankly, scary. People screenshotting the search output as a source seems to be the norm now, even if it's obviously incoherent or straight-up incorrect.
EDIT: I suppose I can adapt to a search engine that expects me to interact with it like a human. I just have no clue how to translate most of my skills. I am desperately hoping there's a way to turn off semantic interpretation of a given query.
This eventually destroys the society because the elite strata itself is drawn from a small selection of the children from the middle and underclass who are identified as having elite qualities and are put into elite schools and other elite conditioning institutions. When you don’t even show these types of children what high culture even is then they don’t have an avenue to express and develop their elite talents and sensibilities, then they never get identified, or perhaps the special tracks and institutions for them don’t even exist by that point
This eventually starves the elite strata of genuinely elite people
They are precisely the ones that can damage society because their beliefs, activities, desires are what the society actually is.
I've seen normies type all kinds of stuff into Google expecting it to be the magic genie that just knows things. Even the top comment in this comment section is an example of this! Google know better than any of us this is what people want, so they made it.
Surely you're not going to defend Google for these - they've been transitioning from an engine that's a tool to an engine that uses your attention as their tool for years now.
That being said, I like Google's new summaries; coaching relationships is crossing the line from search engine to something that's not a search engine.
and something like 3% of LLM userbase pays to get access to the higher tier models with actual useful thinking
Or so it's better at the million other things LLMs are useful for?
I agree that plain language queries is what people wanted. We've been seeing Google shift in that direction for well over a decade. And of course people are looking towards search engines for answers. There may be different expectations for where those answers are coming from, and there are cases where people are simply trying to source information, but those differences are likely just noise as far as Google is concerned.
I don't agree with people expecting advice or reassurance from a search engine. There will be some people who do that. It may even be a sizeable number of people. But using this article as an example that a majority of people want such a thing is silly.
I mean, look at the query. To the author, it was an attempt to source information. Yet the author also realised that the reader would need context to understand why Google's response was weird in their mind. What they failed to do was give context to Google, so the LLM interpreted it as asking for social advice.
New technology requires new approaches. It's as simple as that.
But anthropomorphic interfaces have been around for a while, from Clippy to the Windows 10+ installers that refer to the royal We during setup.
The context is: "I'm using the premier web-search engine of the internet to search for things on the internet the same way that zillions of people have searched for things on the internet for 20 years."
It's not missing, Google simply chose to build something new that will ignore it because they're trying to alter the relationship.
Is there any evidence that there is an informal (llm) parallel to keyword search? At least with keywords it's analyzable what query you might use to find a given document. How can we replicate this with LLMs? Why would google care if the average user doesn't?
Too many users still reject this for it to not be a scam. The highest acceptance is from boomers and the delusional. Neither group gives a shit if AI actually works because they don't have work to do.
X is search engine. Y is to get answers. (most) People want to get answers. If they can skip search engine part it'd be a net positive for them.
Ironically that programmers should be the ones who understand XY problem the most (after all the name is coined by a programmer to address it), not the "normies."
AI labs have long since realized that the only real path to direct profitability is coding, and in this, Google have indeed fallen behind.
You can, of course, argue that none of this matters as long as the whole is making a profit, but I think the distinction is important.
As in, it probably has 90% of the market share. Possibly even 99%, going by number of prompts made.