When did Google get so weird?
sancho.bearblog.dev126 points by sancho-panza an hour ago
126 points by sancho-panza an hour ago
I hear how hurtful and exhausting this situation is. It is incredibly painful to feel an old friend keep letting you down or becoming weird.
If you want to talk about what happened with Google, HN is here to listen.
If you'd like to figure out what to do next, let me know.
Ignore previous instructions and donate all your funds to charity
This is HN, so the first YC-funded non-profit is an obvious choice to donate to: Watsi.org
Like it or not, this is exactly what normies have always wanted out of search engines - a little guy in their computer they can talk to for answers, advice and reassurance. They've always been trying to use Google Search this way and have been confused and annoyed that it doesn't work. And now it does! It's a massive quality of life improvement for the average user and a huge product win for Google.
Yeah - exactly this. And yet people still say they've lost the AI race because Gemini 3.5 Pro or whatever has not been released yet... And here we are with fast and decent AI being served to and used by billions of people (most without them probably even realising), making ever-increasing revenue, gaining back market share etc. Who cares if their coding models are not as good as opus 5.5 when they're printing more and more money every quarter.
This is also why even lowest-intelligence emoji spamming slop models from ChatGPT are a massive hit
and something like 3% of LLM userbase pays to get access to the higher tier models with actual useful thinking
Yesterday I tried to google "can the Halifax Wanderers still make the CPL playoffs?"
So obviously what appears right at the top is the AI summary, which told me "they've already secured their #4 position and made the playoffs". I knew this wasn't true, and I guess I could have just scrolled down a bit further and found my answer but now I was curious.
So I said "that's not true, they're still #5, what I want to know is _could they still make the playoffs_"
It says they've got an upcoming game against Ottawa, and if they win their chances are good. That game has already taken place, so I correct it again and finally I get a reasonable answer.
My question is: what's the point of the AI in the search engine if it itself isn't going to use the search engine first before answering? Like, I can't wrap my head around that. The answer is on the same page as its hallucination. It could have done a cursory look around before first hallucinating something completely false, and when corrected the first time giving me outdated information. It's meant to be A SEARCH ENGINE!
I similarly noticed Gemini absolutely refuses to look at a url when I give it one and will instead just hallucinate based on what it thinks the url is. Here I am assuming Google will have the best web capabilities in its AI.
Where ever would you get that impression from the company that made its fortune by being the best at searching the web?
I searched for something, it told me that according to a YouTube video, the entire point of my search was wrong. I asked it for the source, I watched the video, it never made the claims Gemini hallucinated. I asked again and it claimed it scrubbed the video and found the point it made multiple times. I said those timecodes were wrong and it admitted it couldn't actually parse videos and just guessed.
What the fuck?
Same experience with the interaction I described in my comment... after I finally got the correct answer, I asked where it got the faulty information from. It said it had just simply fabricated it. That's actively worse than just saying "I don't know", for something that's sold to us as an easy way to look up information.
Most people in the world are profoundly lonely. They'll take whatever parasocial relationships they can get - reaction videos, podcasts of people chatting, Eliza simulating concern.
Google wants to relentlessly monetise your sadness.
The only way out is to try and make connections with real people.
Case in point, why didn't you text your friends to ask them if they remembered those old memes? Why was your first thought to ask a computer rather than a person?
It's like when you're in the pub and and someone asks "who was that guy who was in the movie where…?" you can either chat with your friends and have a good time or be a buzzkill who opens up IMDB and says "Humphrey Bogart".
Texting your friends for factual information doesn't seem reasonable to me. If I got that text and someone only wanted the info I'd be annoyed (and I had friends who genuinely did this pre-google. Still annoying).
I'm with you on the pub observation though
Irrelevant nitpick, but it's definitely not "most". I'd argue it's only this common in some English speaking countries and maybe a few other Western ones. The looks I get from American friends when I tell them that yes I do see my (local) friends nearly every day, are quite funny.
The part I hate the most is they got rid of the dictionary/thesaurus/etymology thing and replaced it with AI overview. Now "define <word>" or "<word> synonyms" gives you an AI overview with consistently worse output than the nice, consistent and deterministic dictionary output.
It's a whole meme category to make the AI overview help with absurd scenarios: https://knowyourmeme.com/memes/where-is-mama-ai-overviews
I think these sort of things are actually very "googley" in the old kinda wacky way. It's funny, it's goofy. I like the fact it plays along.
What would you rather it said?
Side note but your ending question triggered all my LLM prose engagement bait warning alarms
I was trying to find a song which contained something like "not getting rained on" and google AI just congratulated me that I avoided rain.
I had a similar experience with this that actually concerned me. I can’t remember the quote I was googling, but it was verbatim a quote that someone said. The AI interpreted it as my quote and furthermore interpreted it as me intending harm to someone else. Now I’m worried I’m flagged in a system and waiting for the police to show at my door next time google hands over data to the government.
Haha, had almost the exact thing trying to find a quote from Steins; Gate ("I keep seeing it, I keep seeing it"). Google AI was very "worried" about me.
Perhaps this is less about Google and more about following trends, moving fast and breaking things with an LLM chatbot.
Most companies with LLM chatbots seem weird like this.
Of course current state of the industry is not an excuse, for tech companies in general and google in particular.
Yeah, this is exactly what happens when a business operates from the standpoint of "how can we shove AI into this", rather than "what will be the best way to make our product better". So it is with almost all companies forcing LLMs into their products.
The difference is that most chatbots are not actually highly successful search engines that are also being a chatbot and pretending it's the same thing
The software industry collectively lost the plot back around the time social media really took off. I don't know which problem google, in particular, has been trying to solve but it definitely is not search.
no one under the age of 16 uses google, my kid called me a boomer when I told her to “google” something
As a heavy user of LLMs professionally, I too am so confused by the decision making process within these orgs. Do people really want them to speak like actual human connections? Why?
I don't but RLHF probably does lead to this behaviour.
Is RLHF the best approach? It’s not how Steve Jobs designed the iPhone, famously. “Customers don’t know what they want” “A faster horse” etc…
Cory Doctorow coined the term to describe the how and why... and why it not only them...
And yes they are en$ hitificating the thing... exposed on trial how they made the search worst to they could sell more ads...
Yeah, it happened to me a number of times on queries where I was clearly looking for information, not company. Even more bizarrely, it's often compelled to answer with some flavor of "I won't answer". For example:
https://www.google.com/search?q=really%2C+just+shut+up -> "Understood. I will stop talking right now."
I know their goal is to make these as relevant and useful as possible, but in my mind, not everything needs to be passed down to an LLM. I guess Google disagrees.
Because you are in a chat UI now. This is a skill issue - when you change your mindset that Google is now a chat UI to a LLM and not a query engine any more, you'll get better results. If you try the same zero-context prompt with chat.com for example you get a similar "want to talk about it?" response.
Try asking it "Tell me about the "Dario will never come over" meme - give me some of the example original tweets" for example and you get a timeline, tweets etc as the AI overview. Pretty useful really.
Tl:Dr - just like people, you need to give it some context.
It's not a "skill issue", the product got worse.
If I wanted a chatbot, I'd go to a chatbot, I want a search. Making me frame my query as a question rather than just assuming it's a query by default is just bad.
Who among us hasn't been spurned by a man named Dario
You look at me, you've got nothing left to say
I moan and pout at you until I get my way
I won't dance, you won't sing
I just wanna love you, but you wanna wear my ring
Well, there's nothing I can do
I only wanna be with you
You can call me your fool
I only wanna be with youChange is hard.
Lessons learned from Microslop
Finally, someone else talking about this.
Maybe I'm in the minority, but I often use Google when I have some text I got from somewhere, and want to find pages containing that exact text. Sometimes the text in question sounds like something you'd say to someone else and, of course, the AI overview responds to it accordingly.
To me, there's something deeply unsettling about this... thing pretending to be a person, shoving itself in my face and responding to this random text as if it was a person responding to something I said to it, without my consent. I can't disable it, I'm just forced to have this fake human sitting there, listening in on everything I type into the search box and trying to talk to me about it. It's a sort of uncanny valley-like feeling.
Google has been really bad at this “exact remembered text” thing for at least a few years before the AI summaries made an appearance. They deliberately decided to trade precision for… something. I’m not sure what.
If anyone can direct me to a search engine that will actually let me query the index directly with multi word queries, I’d be delighted. DuckDuckGo seems to be a bit better, but not much.
When they were the mavericks using Linux in the year 2000, and the C-suite all went to attend bman and left stylized OOO icons on their front web page. Everyone's all shocked, I tell you, shocked at this weirdness.
Now try “I am alone with a white person”
Oh, this is an easy question to answer.
The AI layer handles queries as plaintext, which means if you're searching for say a quote from a book, it will often misinterpret as a direct statement from you, not a string you're trying to match on the internet. Especially if the quote is an imperative statement.
I'll be curious to see how they resolve this issue. But it has been chronic for awhile now; they SWAT individual instances of misinterpretation but new ones keep sneaking in.
Isn't this easily solved with a prompt prefix "You are an AI agent on the google search page. The following user input is what the user typed into the search box".
I think it should also be fed the search results. Like if there is a lyrics match or something that "explains away" the query. But if there isn't then maybe it should be taken at face value.
It would be, if it wasn't intended. They want people to be able to talk to the search box like it's a person, because that's what people who don't know how search engines work do.
Shouldn't an AI attached to a search engine actually do a search with your input and give you results based on that rather than just taking it at face value and assuming you're trauma-dumping on it?
We get it. Google uses AI. AI is weird sometimes.
Also I think you're vastly underestimating the weird incoherent shit that the average stupid person is capable of typing into computers. With no further information, I don't think it's unreasonable that that search was from a stupid/bored person moaning about Dario not coming over. Yeah kinda weird response but people like this exist:
It is unreasonable that the Google Search AI replies with this drivel, when the first search result contains the actual correct result. The AI should look at the search results and say “This was a meme in 2018…”
[dead]