People are being conditioned to think that all the answers come from AI, or that AI knows everything.
But AI is only getting (inaccurate shadows of) the info from these sources which it is simultaneously working to kill. (Websites like this, StackOverflow, human written code)
What happens when there are no more sources and everyone just expects AI to answer everything or write everything? Recursive self-improvement? I doubt it.
Hey, I'm Nicolas from All About Berlin. This post is a few months old. I went through a psychological rollercoaster since I wrote it. Things got worse, but so far, I can still live from the website. Next year might be my last year running it full time.
I'm traveling at the moment, but I'll grab a drink and read this thread later today. In the meantime, feel free to email me. Reader mail would be a lovely break from the flood of LLM-generated outreach emails.
It's unfortunate for 'good' website creators/maintainers but many people would prefer an AI summary of multiple peoples views on "getting a bank account in Germany" generated from lots sites on the internet than clicking through links of content of unknown quality and possibly outdated information that may just prioritize the banks with the best affiliate commission.
To be fair, the website in question is highly recommending banks with an affiliate link first - banks that are in hot waters for violating anti money laundering regulations. If your business model is so dependant on a third party you have absolutely no control over, it wasn’t really a good business to start off with. And yes, LLMs will eat the lunch of any information-only sites that don’t anchor their audiences somehow.
I know of someone who lost their job in journalism, not because they are using AI to write but because the site is completely starved of traffic now.
Long term, with the current state of affairs, why would people continue to create online if it just feeding the machine? Surely we are not going back to paper?
That said, Private Eye never went online and is a successful publication.
The A.I. genie is probably never getting back in the bottle. How I would like for this technology to just not exist.
The hackernews frontpage would be sooo much nicer again as well. I used to browse hackernews being amazed and (pleasantly) surprised about the cleverness and imagination of people. but these days you have to look for the nice stuff with a lookingglass between the onslaught of breathless AI stories.
I just can't be bothered anymore to be amazed about what AI can dow now. I mean, it is amazing of course that we now more or less live in the asimov sf world. But it the kind of amazement I would very much prefer to have stayed a fantasy.
Please may I have some less.
This is the massive elephant in the room; AI needs human-made work, published openly to further develop, and simultaneously heavily discourages human-made work to be published openly, either because the people who previously did it now use AI to do it or because the people who still do no longer have any incentive to publish it openly (and lets be honest, the incentives were pretty meagre before).
But craziest of all is that the best plan this trillion-dollar industry can come up with is "we'll figure something out later". Maybe we will, maybe AI will be able to train off synthetic data or we'll achieve the perfect model that never needs improving or we'll simply hit a wall with how much training data we need, but for a problem that an existential threat to LLM's, that's a lot riding off a few maybe's
I'm working on something[1] similar to All About Berlin but for Hamburg. Instead of going with a website, I went with an app. Sure, it is hard to reach more people but at least, I don't have to deal with this kind of situation and I can create a much better experience.
All about berlin is an incredible resource. Helped me a lot when I moved from the bay area to Berlin and also with all the constant visa procedures post move.
I can recommend signing up to his newsletter for berlin events - https://allaboutberlin.com/newsletter.
People (authors etc) are understandably upset about this. The models CLEARLY trained on data where they didn't own the IP. It just sort of happened too fast for laws to catch up, and now it just is what it is.
But let's assume for a moment that laws caught up. That courts decided that "no, you can't train an AI model by just scraping any content you want, and if you serve an AI model that suggests being trained on such content, you're also breaking the law".
What would happen? This seems like it would be a cataclysmic change to the economy? We'd suddenly "outlaw" several of the world's largest businesses, and the quality of AI models would jump backwards several years (Which on an AI timescale is like a century)?
But what if this is the "right" thing to do? Are we incapable of doing what's "right" now? Did the AI companies' bets on being quicker than legislation pay off?
I respect your work and I can completely understand your feelings. With that, I have but one advice.
Sue.
If they reuse your content, lawyer up and make them pay for it. Nothing will happen unless people stand up for themselves.
The thing about the Google AI overview that I find most baffling is how self-defeating it is. You kill the traffic to sites, the sites die, now you don't have anything for your AI summary to pull from. Also Google is an ad company, they're literally removing their own revenue sources by reducing clicks.
We track the W3C's TDM Reservation protocol uptake, and I'm surprised it's so low, we only see it on 1,113 sites ( last month ), 1,876 sites this month.
https://www.w3.org/community/reports/tdmrep/CG-FINAL-tdmrep-... https://theweb.report/#ai-web
At this point, Google must be aware that:
- They had to introduce AI Search Overviews (competitors would've killed search altogether; now they have some sort of control).
- AI Search Overviews won't work without content creators.
- Content creators won't continue if AI Search Overviews keep eating their lunch.
- They have to come up with an incentive model for content creators to keep doing the things that power AI Search Overviews.
Time will tell.
More broadly, AI is killing the affiliate link and the ad-supported business model. As an AI user I love using AI, because it gives me what I seek (answers) and it takes a minuscule amount compared to what the previous systems took.
While I have never used it, All About Berlin seems like an exceptionally good resource, not showing ads, but rather making its money by driving traffic with good information and having partnerships with affiliate links.
Other than this, most of the ad-supported web I greatly dislike. Usually the noise drowns the signal, because websites competed not on the goodness of information, but the SEO rankings. I have been blocking ads since 2005 and I never used affiliate links either, so from my very narrow individual perspective, I was never a valuable visitor in the first place.
It would be great if the quality sites like AAB could get another stream of revenue that makes them sustainable, even in the AI age. I'm afraid they run into a similar problem to monetizing open source software. At the end of the day, if the information is out there, people and corporations will take it, and not contribute anything back. That won't change, so it needs to be factored in somehow.
Switched to DuckDuckGo as my main search engine some years ago. Not even for ideological reasons, the results were just better.
I have to say from the bottom of my heart: Thank you for making my 4 year stay in Berlin more bearable.
To be fair, Google’s been suffocating search for years, before AI. Actual search results have been pushed below the fold for a long time, if you happen to not use an adblocker.
This is one of the reasons why I try to ignore the AI overview and go straight to the links below.
Recipe sites too. The ai summary is usually a 100% copy of the top recipe result in the links, but stripped of extraneous info.
I unashamedly place 100% of the blame of this on people who continue to use AI. Stop. It is ruining everything.
Everything you do is a choice. They say they'll fire you? Will they? Have they fired anyone for refusing to use it?
They can't fire everyone. Stand up and be counted.
To be honest I'm guilty of using AI for many things I used to use websites for. News (with web-searched sources and citations), Recipes, Exercise routines, Managing my emails
Why? because its a better user experience. It's that simple. News sites in particular (but websites in general) are ad infested hellholes or paywalled to the point of uselessness. An AI app is a single source of pretty much anything I need, for a relatively small monthly subscription and no ads. I can even have claude just build an artifact if I want the data rendered in a nice readable website
If traditional web wants to compete it needs to be a better product first and foremost. People need to pick it over an AI app
Can't people use cloudflare to block AI bots? It has a setting and google says it respects it if one does not want AI training.
Happening to me as well at BookDNA.com :(
Super frustrating to see Google crawl all our hard work, and then stop any traffic to us.
Never blame the technology. Focus on the people, or in this case the corporations wielding that power. It is in their best interest not to cause an extinction event of the ecosystem they thrive on. Currently they (Google etc) are merely leaching on the ecosystem without giving anything in return (destructive). Ideally, you reach symbiosis where there is a net positive result for both sides, ie. enough money flows back one way or another to the content creators you depend on. Either way, eventually this will come back to bite them, but in the mean time the ecosystem, maintained and built by real people, are the ones feeling all the hurt and it's a very sad state of affairs.
Wow at that screenshot. And that's what everyone is using now? Surely it can't be that bad on every search.
We need a new internet that can be protected by LLMs reading or writing it. And in general find a way to isolate the world from big tech by creating competitive products. I feel like if a lot of talented people got together we can do this.
Wecome to the club. I managed a somewhat successful community website from 2003 and by 2018 it became a shadow of itself as users flocked to Facebook, Instagram and the likes.
“Local and federal government offices refer people to my guides. The Berlin immigration office uses my data about themselves.”
Have you tried pursuing grant income?
Google Ads showing Deutsche Bank and Commerzbank at top, two of the most bad guys in German financing! Involved in serious stuff regarding human rights or Russia. Fresh Hacker leak 2 days ago shows connections to unethical stuff I only knew from Swiss banks.
Best German ethical bank maybe: -> Tomorrow Bank based in Hamburg 100% digital, or -> GLS Bank for those who care about others. ;)
Google was always bad to me so I ignored those ads with joy: a company promotes itself via Google? Then I don’t buy t you. Others could follow.
We have the money. We decide. Then we might help those content creators/ Webhosters too.
Dead Internet Theory on steroids in here.
World is changing. Websites once killed the need of magazines and many types of books too..
I rather look at it as an opportunity to demonopolize Google as I do exactly the same (overviews and insights) I just find Google Gemini obsolete as I find better AI solutions where I can check for knowledge and information.
Open Web has a lot of content that was crawled before and I really hope that all that SEO/Traffic hunting nonsense stops to benefit Google towards major regression. They are so 00s company and we're heading towards 30s of XXI century.
Web is dead anyway.
Is the AI generated content as up-to-date as the source site, because if it’s not, we’re all losing out. Is it only as relevant as the last model?
It is nothing new in this case. It hit others with updates to Google's search results before.
The problem is, he is dependent on Google's traffic. This is a big weakness in his business model.
Looking at non-German immigration to Germany (so excluding German citizens returning). Immigration absolutely peaked in 2022/2023… over 2x the 2021 numbers.
On paper immigration in 2021 vs 2026 and the site usage is just slightly down from 2021. Could at least a partial cause be there are fewer prospective and fewer recent immigrants?
Just a counterpoint. I’m sure AI is also a cause.
The web before AI wasn’t all sunshine and rainbows. So many websites were stuffing their pages with filler text to be ‘SEO-friendly,’ and the top results you clicked were full of affiliate links, unnecessarily long paragraphs, and, in some cases, misleading information.
AI has many problems, but "I can not make recurring income from being a digital librarian" is not it.
> It’s hard to fund my work with only a quarter of my audience.
As someone who thinks that all the big issues of the digital economy are rooted in the misaligned incentives between consumers and producers, I think this is a great thing.
> The Berlin immigration office uses my data about themselves.
It's not your data, you just compiled it.
Plenty of people are doing that, for free, on Wikipedia.
Oh, how naive. Did we ever, think that those mildly gatekeeping knowledge (search) would not - at some point - go all the way.-
The Verge has been calling this Google Zero for a while now. Let's see what Sundar Pichai has to say about it:
https://www.youtube.com/shorts/P6kDTC--PGw
Truth be told, I don't quite know what Sundar said here but it may be McKinsey Speak for "suck it up".
Yeah why spend effort in creating something when it just gets stolen and never read by a human?
I am sick of Americans not enforcing anti trust. Americans effectively get to do whatever the hell they want by enforcing a “free market” almost worldwide. A game they easily win as its rigged in favor of the richest country. Subsequently they refuse any anti trust for their tech “hyperscalers” allowing them to monopolize enabling insane margins. Then they pat themselves on the back and proclaim “American Exceptionalism”.
Also the irony of toolcalling is that the AI now uses search to be more accurate. Works great until it cooks all sources of truth. Wonder what the long term strategy looks like
"Put simply, it shows immigrants how to adult in Germany their new country."
What? I suppose "Put simply, it shows immigrants how to adapt in Germany, their new country." is meant.
For the love of everything holy, people, please stop using spell check tools. One or two characters off is i) easy to spot and ii) the intended word usually obvious (at least it usually will be obvious that there is a typo). A helpfully corrected, but wrong word is more difficult to spot and makes the intended meaning hard to guess.
Let me give another example from Barcelona. We were there and with no ticket two months in advance, we couldn't see Sagrada Familia. This is according to all easily accessible information at Google and with Claude researching.
Your only option is to pay weird middleman 10x the price for non-existent or shady tours.
Guess what - it was all wrong. Go there, buy a physical ticket for 2 days later. I think this is all going in the wrong direction. People with malicious intents fake the online data until nothing there is correct.
We saw this enshittification of social media already where it is all far-right now - both ownership and propaganda. People fell for this a few times, but reality hits eventually.
Asking sincerely: if it’s not about money and just about helping people, what better way to help people than to have your content be trained by AI?
Simple. Assume that you know one trustworthy website. It has a bunch of links to other websites that are similar and also verified by the website you're currently on. These websites of course want to be friendly to they link back. But they also link to other websites. In this way, you can discover an entire web of pages that are verified to contain quality material, by going through relevant topics like "travel” -> "travel to Germany" -> "expats in Germany" -> "expats in Berlin".
I have a strong feeling that someone has already invented this technology, but I cannot put my finger on it.
It's easy. Add a license to your website that allows indexing but not LLM training or LLM access. Sue to the ones who infringe.
How do you all feel about the situation?
Is it just okay that these humongous companies just took all our work without paying us a dime?
Is it okay that they are replacing our jobs with tech that would never have existed had they not taken all that work? (or would have moved slower)
Do you feel good supporting these companies by giving them money every month?
What do you think will happen if these companies become profitable? Do you think it will have a positive effect on our society?
Personally I feel that I'm living with severe chronic cognitive dissonance - and have been for many months at this point. But I'm not sure what I can do, I have rent to pay and children to feed.
Do you all just not have this feeling? How do you combat it?