I'm becoming AI-blind

There's some psychological mechanism by which my brain immediately recognizes AI generated text and just short-circuits to "there is no information here".

And when I force myself to read AI-generated text I realize I'm making my brain do creative work to impart meaning to the words. It is exhausting because my brain is literally trying to do a just-in-time rewrite of the text into something valuable.

Something is deeply wrong with AI generated output, and I say this as someone who is typically very impressed by AI.

My son is currently learning Romanian and I was trying to help him with verbs. I don’t know Romanian but recalled when learning a foreign language for the first time it really helped me to break down how a verb form or tense worked in English, then learn the equivalent in the new language. So I wanted to make some charts and pages that he could use as learning resources.

I used Claude to help. I don’t know how to quite describe it, but because the text was polished and well constructed my brain was giving me the the signal “if you aren’t getting this it’s because you’re not focusing” so I’d read it again and then again and it still was not landing. It sorta felt like when you read something technical or heavy when very tired - you are reading but not processing.

Only after wrestling with this for a few days did I realize that it wasn’t me. As I started going through, sentence by sentence, forcing it to re-write things to be more clear the concepts became easy to understand.

I wish there was a name for this situation. It’s almost like a pseudo-language where it has the correct form and presentation but is missing critical components.

The more complex the topic, the more I sense this.

I also find it impossible to parse half the comments that Claude tries to sneak into our pull requests. I’ve never had an issue understanding code comments written by humans like this before. The structure of the information is like a waterfall that leaves me unable to swim to the surface and grab the air of comprehension.

So, “Please write a one-liner comment manually to replace these 5 lines of AI generated comment” is a common refrain in my PR reviews to colleagues.

I've been given AI generated documents or presentations and you can always tell when they've just one-shotted the output. Claude has a way of adding so much jargon and unnecessary text to a document that it makes it very hard to read or even understand what the original intent of the writing even was.

I'm not sure why this issue is so prevalent, it's not hard to point Claude at the Wikipedia article on signs of AI writing or ask Claude to write content anyone of the average American reading level could understand.

To me it just gives off a sense of laziness, that you cared so little of your content that you did not take the time to read it yourself and edit it to effectively communicate the message you wanted to communicate. To that point, it's just not worth my time reading, otherwise my eyes glaze over trying to read between the lines of machine written language for other machines.

That last image is bizarre. The quiche, cream and even the salad look like they've been given the trypophobia treatment.

Which might even make sense, because there were always (still are?) those horrible ads in the chumbox area of news sites that used trypophobia and other creepy body-horror stuff to get you to click. [1] So maybe the hope is that you don't really look closely at the quiche, but some reptilian party of the brain gets oddly activated and drives you towards the restaurant?

1. https://medium.com/the-awl/a-complete-taxonomy-of-internet-c...

These days all I see is, people on the sending side produce huge amount of AI generated text with zero understanding and the people on the recieveing side feed that same text into some other (or perhaps same) AI to scavenge meaning from it. And they do this so much so, that I some time wonder, if we could invent a high density wire/binary format for AI outputs and enable direct agent to agent communication to save some energy and bandwidth.

My current theory is that this reflects a weakness in Claude’s ability to see the “big picture”.

When writing code, I have to explicitly tell it how to structure things at a high level, or the result is sort of a flattened spaghetti. Similarly, when it’s explaining things, it’s not good at pulling out unifying concepts and explaining top-down as a smart human would do. It groups little things together but often doesn’t generalize or synthesize explanatory connections from them.

I’ve been experimenting with explicitly working through a sequence of outputs at different levels of detail, but I haven’t found a consistently successful method.

In high school, a teacher gave me a copy of "How To Read Better And Faster" which teaches you speed reading. This came in very handy in college.

I find that when I try to speed read modern human writing, there are often errors (like missing or misused words) or awkward expressions that I do have to slow down and think harder a lot to really parse it.

With AI writing, it's sort of self redundant and the information density of each sentence seems to have more even information density. This makes it very easy to do a very high level speed read and get the full gist.

There are also what I'm assuming are bots on hugging face (or maybe non-native english speakers who are using ai for translation) that interact with me where I have no idea what they are saying until I read it very slowly.

Claude has become noticeably, painfully worse at writing in the last six months. At this point it’s practically useless for anything except code.

  "There's an ongoing discussion of whether humans are good at recognizing AI-generated text. While most research claims that humans don't really do a good job there, I disagree. "

And I wonder if those of us in tech are the only ones who really care?

I used some check marks and x-es in a work chat, because it was easy to do, and so I could highlight the good and bad outcomes, and someone immediately asked me if I was using AI. I was caught totally flat-footed, because I hadn't used AI, but it looked VERY MUCH as though I had.

I've had the same trouble with AI tech docs, and I struggled to articulate it. There is something difficult about trying to point to the specific "problem" with a given document

I don't like AI-generated text at work, at all. It feels lifeless and unfocused.

But what I hate the most is that it is objectively better than what I had before. No typos, clear structure, and, regrettably, the verbosity and autistic obsession with detail of the LLM is more actionable and useful than the human guy who wrote lists of commands and URLs as documentation, without explaining anything. Or the colleague who writes in uppercase and with question marks and who doesn't make any sense and forces me to engage in an interrogation effort to get to the bottom of what they are trying to say. Or the colleague who simply hates writing--despite being decent at it--and will call you to give you a meandering verbal explanation that lasts two hours of what they want from you. The cynic in me bemoans that we brought this upon ourselves, in more than one way.

> A technical requirements document that describes a rather simple concept in a very verbose way.

This is not AI specific. I have come across many humans who describe a simple concept in a very complex and verbose manner.

Part of me likes the cliche Claude voice. Not because it's good, but because I can immediately recognize it. When I see it in the Claude app/code then it's fine. In the wild it's a sign to me that I shouldn't keep reading.

It may be nostalgia but I feel like we reached peak humanness with GPT-4o and since then it's been getting more and more alien. Particularly Fable.

Idk I’m starting to have have trouble with the frontier models and a couple how to write like a human skills…

I've been a big reader, and many AI outputs nowadays reads polished similar to published books.

The reason I brought it up is because, people who learn English normaly start with a book. It's heavily polished.

When you speak English as you learned from the books, it does not sound very conversational.

If you are native/fluent English speaker, you can feel the impedance mismatch and feel something's off.

The AI-blindness stems from the fact that those polished edits are so common in publishing field, they all sound the same, and unable to recognize the diffs between AI-generated and human-generated.

There is no real human conversational vibe to them and well. i will stop now.

Something that bothers me about AI generated content more broadly is how unmemorable it is. I don't mean as in bad. I mean literally, as in hard to remember or recall.

Despite seeing a lot of them, I cannot think of one AI-generated photo that I can picture clearly in my mind; a few are partial but elusive. Whereas I can recall (visualise) a whole bunch of traditional photographs.

The same is true of AI generated text. Only the annoyances stick. I cannot recall real details of text I have generated, until I commit it to memory some other way.

I don't think this is about ephemerality either. If we assume it's about celebrated/famous/infamous images, there are definitely non-ephemeral, cultural moments in AI generated images in particular, like Boris Eldagsen's Sony Prize winner:

https://petapixel.com/2023/04/14/artist-refuses-prize-after-...

This really should be memorable, but isn't. I had forgotten the second person is in the image.

Or Jason Allen's fake painting:

https://petapixel.com/2022/09/01/ai-generated-artwork-wins-f...

I had already forgotten there's more than one figure in it, and I only looked at it a few weeks back. I remember the colour, the bright circle, some vague hints of texture; one figure. And that is it. Only the crudest shape elements.

For me, something about AI-generated text and images confounds recall. It is really peculiar.

Ironically, the “summary of the situation” linked within this article seems clearly written with AI.

AI;DR

Oh man I can so relate to this.

I can relate. I noticed this exact phenomenon when encountering NotebookLM-generated diagrams recently. Even if I know there's some intelligent thought behind one, it's like some slop detection circuit-breaker is tripped.

[dead]

As people rely more on AI they experience cognitive atrophy. This is measurable in IQ loss, and other symptoms we might otherwise associate with early onset dementia or Chronic traumatic encephalopathy.