Google is overhauling search results with AI overviews and Gemini organization
theverge.com
theverge.com
When I look up something as basic as time zones in two different regions, I don't want AI to imagine something, I just want the results from the thousands of sites available out there to just be presented to me.
The Internet in 2024 is full of not-real things and the effects of financialization. This is the underlying problem, not that Google Search is inept.
Google is the messenger, and blaming the messenger is fine if you want to blame someone but poor if you want a solution.
I have a Gmail account that I use only for purchases. A majority of the mail there is about me spending money. A highly valuable consumer, you'd think?
A far as I can tell, nothing about that account has been passed to any third party, for money or for free.
The operative words.
I'm worried for HowTo websites because they will probably lose a ton of traffic.
If you're searching for a specific site, obviously this won't help, but I would imagine that an increasing percentage, if not the majority, of searches are people just typing questions into the search box. I certainly do that a lot.
This gives you an answer right at the top, and I haven't seen any obvious problems with hallucination yet.
My only fear is how they will monetize this.
1. Google intentionally makes search results worse so that users will do more searches and see/click more ads. There was a big thread about this somewhat recently.
2. Ads have steadly taken up more of the results page and become less distinguishable from organic results. This causes users to more often click on whoever paid Google more, rather than what Google thinks is the most relevant result.
If you think that's bad, just wait until they can use generative video to create reviews of new tech products. (Come to think of it, just wait until they can AI generate video reviews of any products.) A whole swath of people and organizations relying on youtube revenue are going to be taking a 95% chop in pay as Google, or whoever wins generative video, takes all of the youtube payouts.
Knowledge will still be published through papers
People will still write blogs and share ideas for free (As they do on substack which can already paywalled)
IG, TikTok, and YouTube will still be lucrative for creators (live or robot), so how-to videos will continue to exist.
I don't see content going anywhere. Maybe the "10 best restaurants in dallas" but that's fine with me.
For decades now, we had a social contract: Google scrapes our websites and surfaces some of our content, we make it easier for them, Google gets people on our websites, and in exchange Google gets content (links) to stuff with ads (plus they get some tracking on top).
This created an ecosystem where people are incentivised to create better websites. You create good content, Google surfaces it, you get eyeballs, and it's a decent deal overall.
Sure, this dynamic is contingent on Google fighting SEO spam, which they seemingly stopped doing lately, but still there was hope that it's temporary. There was hope that if search gets bad enough they'll start bleeding users to competitors and get their stuff together.
Now Google throws the contract out of the window. Whatever you write or record or snap or shoot, they ingest everything (because we have no choice) and feed it into their Moloch, and out goes unattributed, unverifiable, ad-ridden slop. There is no compensation, no recompense.
Google is midjourneying the entire internet.
Right now I can tell if a website is blog spam SEO garbage, but I'm not sure if that'll be possible once everything is "laundered" into a homogeneous looking result.
I think there's going to be incentive to create huge quantities of "data" because having the highest volume makes you the source of truth for a topic. So, rich companies and people will be buying compute like crazy so they can generate tons of AI spam so another AI surfaces their viewpoint as "fact".
The future is going to suck.
This really is looking more and more like the death of the open web. Curated and closed communities will be the only places left to get authentic information online.
I'm not as worried if that distinction still applies – I can just block 'Google-Extended' and let the normal Googlebot through. If not, though, that feels like the beginning of the end of people wanting to be indexed, eventually leading to Google eating itself.
[1] https://developers.google.com/search/docs/crawling-indexing/...
The problem is, with GPT style AI I think this was inevitable, Google has just decided to ride the wave rather than falling into obscurity.
Also, google has youtube, which should not be underestimated as a source of information. There is a lot of good, credible information on youtube, with even better signals (sub counts) than the traditional web.
This is Googles ace in the hole so to speak, the web can die, they can still feed their AI's with youtube, who else can?
History shows what happens after one party defects on a contract.
Will new human-authored content move beyond the reach of search engines, outside of expensive licensing of social media feeds?
Plenty of AI models cite their sources.
This sounds like a typical overfitting for simple, generic searches, at the expense of anybody who needs precise results. Since google started ignoring operators in some searches, targetted searches have become more and more difficult.
The problem I had was the I'd enter the fridge's model number as part of the search and Google knows it's a fridge, so instead of "exact model disassembly" I'd get results closer to "fridge disassembly". It was useless.
And I felt like I was getting crushed by what I call synonym syndrome. When searching for info about the condensers, the search goes from "exact model condenser not cold" to "fridge condenser not cold" to "refrigerator condenser not cold" to "refrigeration condenser not cold" and suddenly I'm getting results for vehicle air conditioning problems.
It's so frustrating.
I'm pretty sure it does. I'm convinced this AI bubble will pop when it turns out that many of these use cases are not actually profitable. For now there's still plenty of cash left from zero interest era to throw at subsidizing this stuff.
For now these press releases push the stock price up. This is what matters to those corps in the end
I also wonder that....how Gemini and other LLM chat bots decide what to index and include in an answer and what to not include. I guess good old PageRank[0] would be decent here.
More broadly though, it feels like this is going to accelerate the decline of the open web. Fewer and fewer people are going to create quality content as Google steers more traffic towards their own generated content derived from other's original works.
Become happy.
Please note that it isn't a very large index, but it shows what one single person can do today on spare time and until recently a hobbyist budget.
It also very effectively shows that every dumb excuse Google tell us for putting spam on top and not respecting our queries are just that: dumb excuses.
No, I am not talking only about external SEO where some company pays to create links to their page using specific keywords:
This also happens with articles about minor Javascript frameworks that definitely doesn't pay for SEO.
What happens is I search for some less popular (today) framework that a previous dev included, and I get s what seems like interesting results, but when I look into them it turns out many of them are about a completely different, more popular(today) framework.
Undoubtedly someone has put their soul into creating such a system and it is somewhat impressive - but also absolutely terrible when I try to find something and I end up doing unpaid QA work for one of the worlds biggest corporations.
Which is why I pay for Kagi. Yes they use Google and Bing behind the scenes but somehow they have managed to get rid of these massively annoyingly false positives that has destroyed my relationship with Google over the last decade.
And now I can't even use google images without their "highlight something in the image" thing getting in the way.
Any recommendations for a replacement, especially one that considers AI generated plausible looking noise to be spam?
One isn’t even search, it’s generic questions: “how do I make pickled onions” type queries where an LLM can actually make sense of the millions of recipes and summarize them for you. I guess this is the vast majority and what google has been drifting towards for years now as the average user has become less technical.
But we need a completely separate one, based purely on indexing, for “very specific error code stack trace” type searches where LLMs are zero use and current google results are useless. Perhaps an LLM could be trained to separate the blogspam from the actual information to help keep the dataset clean from the inevitable SEO-poison. There’s at last no lack of negative examples in the data…
I recently built a pretty simple static site. I wanted to know how bad the search engines were out to lunch. I chose an industry and domain, then built something simple using a template. Then I did what you should never do with SEO. Keyword stuffing, geo tagging and stuffing, using duplicate content and then copying material directly from high ranking sites in the same industry.
NONE of it was penalized and the site within a week or so was on the top two pages for dozens of specific searches I very deliberately abused SEO in order to achieve.
Nobody is awake right now at wheel. Not long ago, my site would've been buried in a matter of hours. Now? I'm rewarded when I purposefully manipulate the SERPS for some research and get listed on the front page.
If people think AI is going to make this better, they're in for a rude awakening. Its only going to get worse as more people realize how easy it is to manipulate the search engines to get your stuff ranked higher and take advantage of it. There doesn't seem to be any checks on the validity of what the AI is serving up - making it ripe for abuse.
It seems like they have had some system for years already which includes an absolutely unhealthy amount of pages that in their algorithms view are similar to the thing I ask for without being the thing I ask for.
My go-to example is when I search for a construct from a JS framework that a previous dev used and Google gives me lots of results that seem superficially relevant but when I open them I think half or more of the results at the first page are about other, similar frameworks.
This is extremely annoying because they manage to hit a spot were the results are close enough that I often have to read a bit before I realize it.
If Google or its successor can give you exactly the information you need, with references/citations to appropriate sources/alternatives that you can spot check for a sense of security, then everyone is going to use it, and only it.
People that think OpenAI is going to replace, or even meaningfully effect, Google are ignoring basic laws of economic gravity.
But I think you're right. The big winners here will be Big Tech. Microsoft/OpenAI and Google basically.
I own MSFT stock because of it, so yea. I only wish they owned more because OpenAI will be a monster acquisition target.
I'm curious what this will all mean for AWS. Amazon is notoriously behind, so I wonder if we'll see then lose customers because of it.
It seems similar to Meta's bet on the metaverse.
I guess only time till tell if they made the right call or not.
Personally I think it would be better to keep the search they have and A/B test the new feature to see if people like it or not.
The only innovation AI is making from what I see is that it can replace people and doesn't cost anything. It's dumb as bricks and can't say "I don't know."