We went from precise web search with boolean operators to "natural language models" of AI search.
-
@sillyCoelophysis I called my health insurance company with a question about an online order form of theirs that wasn't working. The rep asked me "Do you use Chat GPT?" I said no. He said "Oh well, at least you use a computer." So knowing not to ask an LLM important questions makes me tech illiterate. Wanted to scream.
@fedivergent that's remarkable. How can not using it make you tech illiterate if the point of it is to simplify or democratize technology?
-
@zcutlip @tinker @darthnull @kagihq Sure „On September 4, 2024, Kagi Assistant was announced.[17] The Assistant uses generative artificial intelligence models from third party AI providers“
@lennybacon @zcutlip @tinker @darthnull @kagihq Yeah they do have an AI assistant, but it's not integrated with search results unless you want it to be. You can actually turn off their AI features, and if you ask, their support team will even give you a custom CSS block that hides any mention of AI tools in their UI. That's how I've got mine configured. It's a paid product, but the tradeoff is there are *zero* ads. I've been very pleased with the quality of their search engine. They have a free trial that's honestly worth trying if you're dissatisfied with Google/DDG/Startpage/etc or LLM-based bullshit.
-
@raymaccarthy - I disagree.
I do see your point, but.... I'm not seeing ads in AI search yet (it will happen soon certainly).
The same people implementing the AI search absolutely want as many eyes on ads as possible...
...and right now thats biting them in the ass.
Ironically when folks ask AI a question, THEY ARE NOT going through search results.
Google is getting hurt because folks ask a search question. Read the AI output. Dont go through web site listings or click on web sites.
AI - right now - is HURTING ad revenue.
Edit: I may be wrong here. Things are moving really quickly. I'll leave the above up but read responses below.
@tinker @raymaccarthy that's not what they're reporting.
In Q1 Google's reporting a 15.5% increase in ad revenue, and a DOUBLING of the number of search queries (thanks to a reduction in search accuracy?).
Dare Obasanjo (@carnage4life@mas.to)
AI has improved the ad business of both Google and Meta according to the NYT. Meta’s ad revenue grew 33% in Q1 2026 while Google’s grew 15.5%. AI tools can translate ad copy, generate ad variations, and reduce irrelevant ads by 40%. While better content matching has boosted video watch time (Meta) and doubled search queries (Google). Targeting has shifted from manual selection of demographics to AI-driven recommendations that find the best customers automatically. https://www.nytimes.com/2026/04/29/technology/ai-artificial-intelligence-ad-boom.html
mas.to (mas.to)
-
@tinker @raymaccarthy that's not what they're reporting.
In Q1 Google's reporting a 15.5% increase in ad revenue, and a DOUBLING of the number of search queries (thanks to a reduction in search accuracy?).
Dare Obasanjo (@carnage4life@mas.to)
AI has improved the ad business of both Google and Meta according to the NYT. Meta’s ad revenue grew 33% in Q1 2026 while Google’s grew 15.5%. AI tools can translate ad copy, generate ad variations, and reduce irrelevant ads by 40%. While better content matching has boosted video watch time (Meta) and doubled search queries (Google). Targeting has shifted from manual selection of demographics to AI-driven recommendations that find the best customers automatically. https://www.nytimes.com/2026/04/29/technology/ai-artificial-intelligence-ad-boom.html
mas.to (mas.to)
@sherbang @raymaccarthy - Sure enough! I stand corrected!
-
@AimeeMaroux @tinker
probably undiagnosed
just kidding - the OP was just a generalization, of course there are neurotypicals who dislike bullshit and value preciseness, and there certainly are neurodivergent people who love LLMs (some people describe them to feel like gambling, i.e. dopamine hits when it seems to work, so it keeps you using it, which can be a trap especially for folks with ADHD - and one reason I avoid that crap like the plague)
@Doomed_Daniel might you be able to point me to articles or other sources that expand on this topic?
I have an ADHD husband and he's spending days and nights coding with LLMs. He's obviously having a lot of fun, and there are actual results, but I see the addiction potential, with him losing sleep to vibe coding and getting hyperfixated on his "programming" endeavours, not listening to my less sparkly everyday talk.
I also really need him to focus back on real life from time to time
I'm autistic myself, and I enjoy hyperfocus as much as the next neurodivergent person... But it *is* becoming a bit of a thing.
-
@tinker My favourite notion now is to reject the negative infused neurodivergent and neurotypical labels. I propose :
neurodiverse or neurocreative
and
neurosimple or neuroinert
Anything seen as in the minority is so often couched in negative terms. Often as it is a threat to the rest.
I often find 'normal' people severely lacking in curiosity and creativity and passion.
@NicelyManifest @tinker neurosparkly!
-
@sherbang @raymaccarthy - Sure enough! I stand corrected!
@tinker @raymaccarthy those numbers are strange though. I'd expect a doubling of queries to lead to more than just 15% increase in ad revenue.
Perhaps both are true? They saw a decline in ad revenue so they made the algorithm worse to increase the number of searches to increase ad impressions?
These days I use Metaget, Mojeek, and Qwant for search. I used to use Google exclusively, so they've lost my ad revenue.
-
We went from precise web search with boolean operators to "natural language models" of AI search.
You can tell nuerodivergent, Autistic, ADHD, AuDHD, etc folks created the early internet...
...and you can tell that neurotypical folks are now leading the current overlays of the internet.
We used to have very precise search mechanisms. Specific words found in web pages with boolean operators (AND, OR, NOT, etc) to filter out the web pages that contained specific words and did not contain other words.
Now, we search for web sites (or don't even search for web sites, yay abstraction layers that separate us from actual raw information) using "natural language" to try and coax out info.
Have you ever been frustrated when you use very precise and direct language to communicate a specific idea with someone who then takes those specific words and adds obscure meaning and connotations and personal fears and bias to what you said... thus completely misunderstanding you... only to then try and clarify what you said with more precise language only to have that further degrade the conversation?!
Yeah. That's the internet with AI "search" now.
They took something that worked precisely and directly and muddied it.
We've introduced the "double-empathy problem" to web search.
I'm noticing that everyone in my circles, family, and especially work that are nuerotypical LOVE LOVE LOVE the new AI search mechanisms. They'll tell me exactly what the answer they received was - regardless of whether it's right or has multiple possible and conflicting answers. They just repeat what the AI said like it was the Gospel Truth.
And they love talking to it like it was a "real person."
It certainly takes my search parameters and adds its own interpretation to which I have to clarify and correct which is then misinterpreted further...
I haven't tried vibe coding, but I can only imagine the horror.
Can you imagine vibe pentesting with Claude / Mythos?!?!?!
You know how neurodivergent folks gravitated towards IT because it was precise?
Yeah, that's gone now.
@tinker I'm conflicted about your description there. As an #actuallyautistic person, I totally agree with you and I have the same feelings about AI and "natural language" modles.
However, with lots of #actuallyautistic social contacts and also having been in various self-help groups, I have met a number of violently pro-AI folks who are on the spectrum and don't share my (or your) disagreement about the use of AI as a replacement for actual precise communication with the machine.
So it doesn't seem that being autistic is a key factor in how much you like or dislike the "natural" language that LLMs force on us.
-
@ariarhythmic - No govt law, sure. But many jobs are requiring them (either through explicit rules directing their use or through not hiring folks who dont have experience in them, etc et all)
So. Yeah. Folks are being forced into this in order to survive.
@tinker @ariarhythmic And if you want to do search without AI, there are some hurdles, and the results will contain AI content. So never as good as it once was.
-
After reading some of the replies to the OP (and you should read them, too. It's a good discussion) - I need to clarify.
I'm using neurotypical / neurodivergent (autistic / allistic, etc) as a fairly general kludge. Even quantifying what those mean is currently in flux as we sort out and dive into these ideas.
I have lived experience with two rough groups of people. Folks that use precise and direct language and folks that use language as a way to vibe, feel, and engage with others (rather than communicate specific thoughts, ideas, and concepts, etc).
And I see how one group interacts well within their methodologies but has trouble with the other group and vice versa.
I even know folks from both groups that act as translators and understand each other.
As a quick aside I have two sisters. One I would describe as neurotypical and the other (and myself) as neurodivergent. We all understand each other as we grew up together and we very specifically understand the differences in our communication styles and will even "code-switch" (not a perfect metaphor) with each other.
Anyhow. Take those current terms and understand that they're imperfect...
...which makes that concept of a "natural language" model even all the more hideous to me.
Natural to whom?
@tinker I have a theory that the people who love the AI crap are the same group that never learned to use full text search and was frustrated when no relevant results were returned after typing a whole paragraph of text into Google search, starting with, “so, uh, like, you know”.
What’s worse, a chatbot will often return different results if you ask the same question in a more formal/informal way.
-
We went from precise web search with boolean operators to "natural language models" of AI search.
You can tell nuerodivergent, Autistic, ADHD, AuDHD, etc folks created the early internet...
...and you can tell that neurotypical folks are now leading the current overlays of the internet.
We used to have very precise search mechanisms. Specific words found in web pages with boolean operators (AND, OR, NOT, etc) to filter out the web pages that contained specific words and did not contain other words.
Now, we search for web sites (or don't even search for web sites, yay abstraction layers that separate us from actual raw information) using "natural language" to try and coax out info.
Have you ever been frustrated when you use very precise and direct language to communicate a specific idea with someone who then takes those specific words and adds obscure meaning and connotations and personal fears and bias to what you said... thus completely misunderstanding you... only to then try and clarify what you said with more precise language only to have that further degrade the conversation?!
Yeah. That's the internet with AI "search" now.
They took something that worked precisely and directly and muddied it.
We've introduced the "double-empathy problem" to web search.
I'm noticing that everyone in my circles, family, and especially work that are nuerotypical LOVE LOVE LOVE the new AI search mechanisms. They'll tell me exactly what the answer they received was - regardless of whether it's right or has multiple possible and conflicting answers. They just repeat what the AI said like it was the Gospel Truth.
And they love talking to it like it was a "real person."
It certainly takes my search parameters and adds its own interpretation to which I have to clarify and correct which is then misinterpreted further...
I haven't tried vibe coding, but I can only imagine the horror.
Can you imagine vibe pentesting with Claude / Mythos?!?!?!
You know how neurodivergent folks gravitated towards IT because it was precise?
Yeah, that's gone now.
@tinker amazing post. Yessss.
-
We went from precise web search with boolean operators to "natural language models" of AI search.
You can tell nuerodivergent, Autistic, ADHD, AuDHD, etc folks created the early internet...
...and you can tell that neurotypical folks are now leading the current overlays of the internet.
We used to have very precise search mechanisms. Specific words found in web pages with boolean operators (AND, OR, NOT, etc) to filter out the web pages that contained specific words and did not contain other words.
Now, we search for web sites (or don't even search for web sites, yay abstraction layers that separate us from actual raw information) using "natural language" to try and coax out info.
Have you ever been frustrated when you use very precise and direct language to communicate a specific idea with someone who then takes those specific words and adds obscure meaning and connotations and personal fears and bias to what you said... thus completely misunderstanding you... only to then try and clarify what you said with more precise language only to have that further degrade the conversation?!
Yeah. That's the internet with AI "search" now.
They took something that worked precisely and directly and muddied it.
We've introduced the "double-empathy problem" to web search.
I'm noticing that everyone in my circles, family, and especially work that are nuerotypical LOVE LOVE LOVE the new AI search mechanisms. They'll tell me exactly what the answer they received was - regardless of whether it's right or has multiple possible and conflicting answers. They just repeat what the AI said like it was the Gospel Truth.
And they love talking to it like it was a "real person."
It certainly takes my search parameters and adds its own interpretation to which I have to clarify and correct which is then misinterpreted further...
I haven't tried vibe coding, but I can only imagine the horror.
Can you imagine vibe pentesting with Claude / Mythos?!?!?!
You know how neurodivergent folks gravitated towards IT because it was precise?
Yeah, that's gone now.
@tinker I miss when search worked
-
We went from precise web search with boolean operators to "natural language models" of AI search.
You can tell nuerodivergent, Autistic, ADHD, AuDHD, etc folks created the early internet...
...and you can tell that neurotypical folks are now leading the current overlays of the internet.
We used to have very precise search mechanisms. Specific words found in web pages with boolean operators (AND, OR, NOT, etc) to filter out the web pages that contained specific words and did not contain other words.
Now, we search for web sites (or don't even search for web sites, yay abstraction layers that separate us from actual raw information) using "natural language" to try and coax out info.
Have you ever been frustrated when you use very precise and direct language to communicate a specific idea with someone who then takes those specific words and adds obscure meaning and connotations and personal fears and bias to what you said... thus completely misunderstanding you... only to then try and clarify what you said with more precise language only to have that further degrade the conversation?!
Yeah. That's the internet with AI "search" now.
They took something that worked precisely and directly and muddied it.
We've introduced the "double-empathy problem" to web search.
I'm noticing that everyone in my circles, family, and especially work that are nuerotypical LOVE LOVE LOVE the new AI search mechanisms. They'll tell me exactly what the answer they received was - regardless of whether it's right or has multiple possible and conflicting answers. They just repeat what the AI said like it was the Gospel Truth.
And they love talking to it like it was a "real person."
It certainly takes my search parameters and adds its own interpretation to which I have to clarify and correct which is then misinterpreted further...
I haven't tried vibe coding, but I can only imagine the horror.
Can you imagine vibe pentesting with Claude / Mythos?!?!?!
You know how neurodivergent folks gravitated towards IT because it was precise?
Yeah, that's gone now.
@tinker shiiiit....*this*...
-
@tinker I'm conflicted about your description there. As an #actuallyautistic person, I totally agree with you and I have the same feelings about AI and "natural language" modles.
However, with lots of #actuallyautistic social contacts and also having been in various self-help groups, I have met a number of violently pro-AI folks who are on the spectrum and don't share my (or your) disagreement about the use of AI as a replacement for actual precise communication with the machine.
So it doesn't seem that being autistic is a key factor in how much you like or dislike the "natural" language that LLMs force on us.
@hzulla @hzulla - Yeah, I think you're right. I followed up the post with this one: https://infosec.exchange/@tinker/116496096417501652
I have observed different groups that line up with this pattern and its *often* (but not perfectly) along neurodivergence.... But!
You're right. One guy I know who is nuerodivergent and he is massively into vibe coding.... such as it is.
So I don't know the right words to describe it yet or the exact phenomena or reason for it. I just see a distinct pattern and have found others that observe it too.
The discussion following this post has been good to further flesh it out. I appreciate your response and input!
