FloraCat
@FloraCat

There is an existential dread caused by ChatGPT. We can see the power that deep learning is exposing in generating new content. The content quality is reaching the point where we are starting to feel that it is difficult to distinguish a post from ChatGPT form that created by a human. This is a kind of one way Turing test, where we read the content and guess if the content is intelligent or not.

The dread comes from what comes next. Obviously ChatGPT is an amazing producer of Captain Picard fan fictions. We could imagine a world where all of fan fiction was generated by an AI rather than by people. We could have endless unique Captain Picard fan fictions or unique SCPs only found by the person giving the prompt to the AI. But no on is afraid of losing their job because an AI is writing a fan fiction for them.

Dan Olson recently made a documentary on the Mikkelsen Twins. In it he explored the content of their course on producing books for sale on Audible. The process involves finding a topic that would generate enough interest, and then hiring a ghost writer to string along to write a book. Then finding voice actors to read out your shallow poorly thought out book and flogging it on Audible in the hope that someone buys it by mistake. Dan attempts to write a book using this technique over a month. It's an incredible struggle, but he does manager to accomplish it.

Part of the process that limits the creation of such works is the human sweat effort that is involved in the creation of such works. Someone has to sit down and write all the words that fill the book, they have to be at least logical enough to pass an obvious sniff test to get published. There is an invariant limitation in how much work could be procured in this way. At some point he works are not profitable and that is the cut off point for their creation.

Here is the existential threat to the world that ChatGPT creates. If the effort requires to create a work was reduced by orders of magnitude. We are going to lose natural bottle necks in the world. These bottle necks limit the amount of content that is created and shared. The natural bottle neck in this example is the time and effort to produce an audio book. I'm speculating that ChatGPT can reduce the cost of ghost writing, there is one bottle neck on the quality of voice acting. If that was removed, the flood gates would be opened. A book could be consumed by as few as a single person to be profitable to publish in this future.

This should be an existential fear for Amazon. I'm sure they are already working on ways to fight the inflow of future spam books. I'm sure they already have ways to reduce it before this flood gate was opened up, it's just now that we are seeing the flip side and speculating on it.

It's not just ChatGPT, this has been an existential worry for a while. People have been talking about bot armies on social media for the last decade. The arms race between AI detecting bots, and AI creating smarter bots is going to rage on for the next decade.

What the big fear is for us humans in the middle. Humans have only recently managed to pull ourselves out of world of illiteracy. And now we have the hoards of barely literate people arguing with each other over the internet.

Each and every one of us are going to have to fight more and more with our own version of the Turing tests. Learning to spot good reasoned arguments from just ones that are started with confidence. We are not ready for this world. I'm not ready for this world.

The internet has spent the last 25 years becoming a large amplifier for ideas. It has been wonderful in connecting with many folk that know more than I know, and them sharing their experience with us. We all have better access to high quality information, information that is obscure, and niche. The Internet has exposed it to us all.

An amplifier doesn't just amplify signal, and the noise of the internet has also been amplified. We have had feedback in our minds from this in-group outgrip dynamics. The noise and misinformation that makes people think there is a Queen of Canada (not to be confused with the Late British monarch).

I don't use tick tock myself, but watching my partners 60 year old dad fall asleep listing to military propaganda from both sides of the Ukrainian - Russian war is terrifying. Content I'm not entirely sure was actually crated by a person. Propaganda crafted by a person or a machine is irrelevant. It is noise being broadcasted into the minds of our peers.

It's a well know debate strategy that it takes more effort to rebuke a topic that is raised than it is to raise the topic. So it makes sense to raise dozens of topics and know that by the time they are done rebuking a few, they lose because they didn't rebuke them all. Noise wins over reason.

This has gone a little longer than I have wanted. ChatGPT looks to me to be the next stage in Internet noise amplifiers. The 1st being the many-to-many connections that social media and the internet enabled. This will piggy back on that many-to-many net and increase the amount of illegitimate actors.

We don't need more noise amplifiers, we need filters. And filters are harder to create.


You must log in to comment.