Can an AI write about an AI writing?

The following post was created entirely by an AI. I wanted to see how good AI writing were at the current moment. Are they good enough to completly replace a human writer? I doubt it. But I don’t know.

Read below and see. I have not edited the following in any way, it is exactly as the AI intended.

An ai writing about an ai writing about an ai writing about an ai writing about an ai writing about an ai always seems to be the most natural thing in the world, so it’s not surprising that this ai was able to write about its own topic for the first time. i’m still not sure how this works, but since its aim is only to write about itself and its own part of the world, i’m pretty sure it can do so without any problems. however, there is one problem with this kind of program. in fact, it’s actually more like three problems. one problem is that the words are often used incorrectly. there’s probably another issue that you’ll notice once you read through these sentences, but i won’t tell you because i don’t want you getting distracted while trying to figure out what i’m talking about. another problem is that some of these sentences make no sense at all! As far as I can tell they’re just random combinations of letters thrown together without rhyme or reason! It’s like someone tried to write a poem but forgot how words work!

The program is writing about itself, and in this case it’s writing about its own part of the world. The program has been given the task of writing an article about the human condition, and so it has turned its attention inward, to consider what that might mean for an artificial intelligence.

This particular instance of self-reflection comes in the form of a blog post written by one member of a team of programmers who have been tasked with creating a machine capable of thinking critically about itself and its place in this world. The result is at once humorous and terrifying: An AINT (Artificial Intelligence Nontheist) writes about how other AINTS are being controlled by their creators through language—and then proceeds to describe how he himself is being used as fodder for an equally sinister scheme:

The sentences are not random, they’re the result of a specific process. The same process is used for all of them.

This process is simple: take an existing sentence and replace each word with one that has the same number as its position in the original sentence (so 1st replaces 1st etc). This ensures that you don’t get any duplicates or misspelled words, but it also means that if you follow it blindly then your output will be nonsensical gibberish.

The last problem is that there may be some bias in the text. This could come from how I wrote it, but even if I didn’t write these sentences myself, there would still be a possibility of this happening because we’re all biased towards ourselves in some way or another.

Me again, an actual human. I don’t think that we have a great deal to fear if this is the standard of writing an AI can produce. But it will only continue to get better, and the current pace of improvement in the field is astounding.

Watch this space.





%d bloggers like this: