Vale Wikipedia

This is a post in response to a kind of mental chafing, where a thought gets lodged in my brain and gets me irritated, so I'm purging it here.

I've never had no money, I've never struggled so much financially that I've failed to donate to the wikimedia foundation. Wikipedia is so fucking great.

I also remember how it was treated:

'Tis but a sketch, from Collegehumor no-less, published some time in 2009-10. But this was a popular meme that went around when Wikipedia emerged as one of the internet's most visited sites.

In some ways, when wikipedia was a story, it was just archetypal. Like it's essentially the same thing as this low-rent Conan sketch about Google's attempt at a self-driving car. Similarly, I can recall going to Sustainability events around the same time, and people would talk about creating a global wiki-government through grassroots citizen participation to address climate-change and other utopian optimism much like some people talk about "AI" now curing cancer and solving climate change etc now.

What was different though, and what is so irritating, is that people disparaged wikipedia as unreliable, and wikipedia was great and cheap and free and ad-free, and while just about every other contemporary platform has gotten worse from Google-search to Amazon to Facebook etc. Wikipedia remains terrific.

ChatGPT puts wikipedia, and reddit and newsmedia and whatever else into a blender, then uses an incredible amount of resources, something like 10x or 100x the energy of doing a google search and then just clicking on the wikipedia link. Gemini produces for you, a worse and less reliable result than the wikipedia article, and yet...

For me a literally unfathomable amount of money has been poured into promoting just this application of generative LLMs. That's the big difference, using a little hyperbole license, it's fairish to say nobody invested in wikipedia in any financial sense. A community produced a social good, which was great because it updates constantly, but is also checked constantly to enforce community standards. ChatGPT 4.whatever now has a crappy little message at the bottom to disclaim that the content may not be acurate. If you go to a wikipedia page on a subject that few people are interested in, there will be a big box up the top warning you that the article has numerous problems with it. Claims will be followed by "[citation needed]" there's a talk page where you can see conspiracy fiction fans demanding answers as to why the subject of a page has to be introduced as a "pseudo-scientist" and "conspiracy theorist".

I can only speak for myself, but my direct experience of the ease of adoption of Wikipedia into my life vs. the friction it was given by the media and contrasting that to the continual attempts to force me to adopt "AI" features from companies burning money to try and fuel this revolution and with literally pretty much every business exec in the world desperate to believe it will soon help them reduce headcount and hyperscale their output.

It's ridiculous. It's a ridiculous time to be alive. I don't think generative LLMs have no use cases, I don't think it will, nor needs to be "uninvented". I think the returns on investment just aren't there and it vies with Private Credit as a herald of the next GFC-like market collapse when we have enough data now, that we are really just waiting for silicon valley to admit the "AI revolution" is just the latest in a 12~13 year series of dud investments, but now interest rates aren't 0% they need returns.

Hail to Wikipedia. The Queen ain't dead yet, the rest of the internet just about is though.