💾 Archived View for the-brannons.com › hype-train captured on 2024-05-26 at 14:51:07. Gemini links have been rewritten to link to archived content

-=-=-=-=-=-=-

Take a Ride on the Hype Train

moddedBear poses the question, is AI this year's metaverse?

VR has been hyped before and it doubtless will be again. AI? Same deal. Tech companies have been overpromising and underdelivering for decades. Nothing new under the sun. AI isn't this year's metaverse; it is this year's social media.

Metaverse was Facebook's gamble for relevance post-2020, when a lot of people woke up and started realizing just how terrible Facebook was. It was a blatant attempt by Facebook to make itself the center of a new "reality". Nothing more, nothing less. I'm not surprised it fizzled. And I'm still not quite sure what a metaverse is.

The current AI hype is a bit more ... organic. This stuff's been cooking for years, and it has a lot of nerd cred. People are finding new toys and playing with them, as people are wont to do. Rather than an attempt to create a hype train using a throw of the dice by a terrible company, you have companies jumping on a hype train that was already rolling.

One thing is worth noting about business. It's very herd-like, or very lemming-like. If Johnny down the road is jumping in a lake or dying in a fire, then I've gotta try it too! Because Johnny is one of the cool kids. Concentrate all of this power and money in one place -- Silicon Valley -- and what you do is amplify the herd behavior.

I'll also note that big tech usually doesn't actually innovate. They sometimes make improvements to existing things, but real innovation is rare.

Take Google. There was search before Google. Google did it better. I'd probably describe Page Rank as a real innovation. They packaged it well. In the late 90s / early 00s, they gained nerd cred by having a very spartan homepage. A thing that delivered results without fluff and bullshit.

But what did they do after that? Google Earth? Microsoft was there first, if I recall correctly. There was a Microsoft thing that people were gaga about in the late 90s, and I can't even remember its name.

Gmail? There was Hotmail and Yahoo. Google had better spam filtering and more nerd cachet in the middle of the first decade of the 21st century. And OMG, all of that free mail storage space! None of this is innovative.

Android? They were following Apple's lead. And Apple wasn't even first. Apple just had good marketing and a cult following, especially among segments of the elite.

Google Plus? Yet another attempt to follow the leader, when social media was yesterday's AI. It was a boondoggle.

Now for some predictions. We have Google and Microsoft trying to integrate GPT with their search engines. That's going to fail hard, because GPT is basically just a very expensive bullshit artist that uses a lot of computing cycles. I'm sure there will be other GPT-powered huge failures. The hype will fizzle, but the technology might improve as we learn from failure. It'll go back to "cooking", once some big passengers fall off of the hype train and get smushed against the rails.

What GPT and VR have in common right now is that both are attempts to turn CPU cycles into believable illusion.

PS. The kind of AI that turns me on isn't the generative kind. Of late, I've been playing with whisper.cpp, using it to generate transcripts of talks and whatnot. This kind of thing is a human amplifier, especially when it can be decentralized. There's a lot to be said about the difference between computers as human-amplification tools and computers as babysitters / experience-providers. I want a tool, not an experience.