2 min read

Artificial Intelligence Hellnado

Artificial Intelligence Hellnado

I was drafting an email to customer service and I was stuck.

I opened ChatGPT and wrote:

write an email to a customer service agent of <redacted>. Explain that you have enjoyed the product for several years. You are currently having trouble adding a <redacted> credit card account to your account. You opened a chat and followed suggested troubleshooting steps to no avail.

What I got in response was a passable email that I edited. I wanted the email to read "I'm friendly, but I'm a little concerned."

This whole experience reminded me of a service I came across before ChatGPT really caught fire. This service aims to mitigate the power asymmetry that characterizes the majority of our interactions with large corporate and governmental entities.

The founder, a Thiel Fellow so you know he is a freak libertarian nerd, built a product that uses AI to fight everything from parking tickets, to overdue fees, and airline injustices. He envisions a future where we all sick AI bots after each other's AI bots. It's pretty grim, but then again, I'd prefer to avoid customer service calls as much as possible.

Annoying nerd Ezra Klein put out a podcast episode about advertising. The guest discusses that the real innovation at Google in the early 2000s was the realtime advertising auction.

When you search, Google announces "This person with A and B characteristics just searched for X phrase. We'll start the bidding at $0.01." Next a bunch of advertising firms place their bids and the winner gets to put their ad next to whatever you searched for. The transaction for you attention all happens in the time that it takes Google to return your search results.

Imagine an extra step in the above auction where the advertising bidder takes the auction information and feeds it into something like ChatGPT:

Write ad copy to sell my dumb thing to a person with A and B characteristics who just searched for X phrase.

Imagine further that the advertiser takes the same information and sends it to an image generating AI, like Midjourney.

Make a photo of my dumb thing that a person with A and B characteristics who just searched for X phrase will click on.

We also send that information to a video AI to make a Youtube ad.

All of this happens in the time it takes to click with your mouse or scroll on your phone.

Very cool nerd Jaron Lanier often talks about the internet as a Skinner Box. A Skinner Box is an apparatus used to study animal behavior. Lanier means that big tech companies know a lot about our behavior and can supply feedback designed to shape those behaviors in realtime. The more behaviors that can be captured and measured and the more realtime the feedback, the more perfect the Skinner Box becomes at controlling the occupants behavior.

We can imagine increasingly elaborate AI prompts that produce better and better results to shape all of our behaviors on the internet. This can happen now.

Lord nerd Neal Stephenson in a recent book wrote about a near future where the various content feeds on the internet are completely over run by insane garbage. In Stephenson's near future, the wealthy are able to pay for human curation of their feeds. The non-elite, who can only access unfiltered content streams, are mutated into extreme caricatures of the most red-pilled Trump supporter you can imagine.

That doesn't seem too far off from where we are. Our feeds are already becoming increasingly overrun by AI generated content. The wealthy can subscribe to various newsletters.

Write a post about artificial intelligence in the voice of a very normal nerd.