All the information Gemini surfaced was created with human effort and published on the internet with the expectation that humans would visit the website and the creator would get some reward - advertising dollars, bragging rights, popularity, subscribers or whatever else.

If the only visitors to websites are now LLM training bots then what incentive is there to publish anything new? For how long can we continue to rely on pre-2024 non-AI generated content?

That works now because there are human made sources that the AI can find and summarize for you. But now there are no incentives at all for humans to write anything on the internet and if they do the content will be buried by hallucinated content someone else posted at a larger scale.

I had the similar experience to yours yesterday and it lead nowhere. Funnily enough I was also trying to configure a vpn on a router, google didn't return anything useful (besides a blog post clearly written by AI and with absolutely no information in it). Claude managed to give some interesting pointers, but its suggestions were not working and I also noticed that it started to hallucinate badly about ipv6 and gave me some suggestions that were just plain untrue. Claude Opus is smart, usually when it gets so convinced about something is after researching the internet and not just based on its training data. I wonder where it got so convinced about it. Maybe reading some other hallucinated blog post like the one I stumbled upon?

That only works when there's an abundance of documentation for your specific device. The moment you're on a more recent version of something and have a weird issue, all hope is lots. You get stuck in loops because all LLMs keep recycling old advice that no longer applies. This problem will only get worse and worse as people no longer as questions on public forums, so answers are not publicly available either.

Yes. When my kids wonder why I don't hate AI the way they do I tell them it's because I hate Google even more.

To be more precise, I hate the SEO shithole the internet has become, that Google serves up, that Google facilitated, indirectly created.

(I really don't have any tears to shed if there is a death of the Corporate Internet™.)

While I understand that you feel good as you got the device configured, you would have been better off without gemini. The 'old' way as you call it would have resulted in you knowing the backgrounds and the inner workings of your router, making maintaining it a breeze and helps you actually understand your setup. Besides that, it would probably trigger you to rethink some of the things you now blindly have implemented because gemini did not show alternatives nor reasoning behind it. (something that most definitely would have been documented on the source pages)

Yep - it's awesome, the problem is that Google isn't sharing the revenue with the context creators anymore - over time, unless fixed, this will decimate the knowledge base it feeds on.

> Oh, I should mention though. There was no advertising at all. They didn't make any money off me

Are you sure about that? Even if you didn't see ads ( remember people pay even if you don't click - just like a billboard ) - they are still profiling you to better sell you ads in the future, and using your interaction as free training data.

I've run into major problems with LLMs as I maintain my home Linux systems. If I just copied commands they list, I would have a near 100% failure rate, as most of the information they have has been gleaned from forum posts that are years out of date.

They're a good jumping off point, but I need to delve into the original sources just like I did when I used Google.

I hate to break it to you, but we are at the start of enshittification circle. Once Gemini has monopoly you will reminisce with joy google search results.

I've been mostly enjoying Gemini, but it also clearly and definitively told me something i was trying to do was not possible with the library im using, so i wrote a different implementation, an hour later to discover that the library does in fact do precisely what i wanted in exactly the way i wanted with less headache. If I'd just gone right to the documentation instead, it actually would have saved me time.

Believe me they're making money off you one way or the other

And I've spent the last few days irritated that everything I ask Gemini is answered with something that's blatantly wrong and I'm not even a subject matter expert. A quick Google search for the same questions gives plenty of results that counter what the LLM gave me.

YMMV.

its pretty good with logs too. i mostly paste a few pages i suspect hace a problem in them and let it go to town. its almost always correct and is way faster than me

You deserve what's coming for you.

I built a simple Claude Code container on my homelab to do the same. I can just SSH in and get tech support when I need it.

AI is on the whole bad but it’s impossible to forget that the ostensibly human-curated Web as presented by modern search engines is bad too. Invasive ads, popups, and worst of all the substantive content has a nine inch frame of SEO filler and a three inch picture (what you were after). And some things are not even high-tech slop stolen. It is just old-school verbatim copied from another website and repackaged with another frame.

But yeah, text remix machines are not a long-term solution to that problem.

Imagine someone making an indie router and sell it too you then the AI not be able to surface their doc

no, no, no. you are supposed to romanticize the hunt for correct information /s

There will be new ways and incentives for content creators to be compensated. Many AI search startups are already talking about this or have created programs that help incentivize content creation.

Are you sure the lack of advertising isn't long term feasible? It seems to me that AI models have proven themselves to be something consumers ARE willing to pay a subscription - or even pay per use/token for.

It took you 4 days because you used Gemini. Gemini is the worst AI model I ever used. It is way behind even open models. It looks like Google just reached its AOL moment.