Why AI writing is so generic, boring, and dangerous: Semantic ablation.

[cstross@wandering.shop]

Why AI writing is so generic, boring, and dangerous: Semantic ablation.

(We can measure semantic ablation through entropy decay. By running a text through successive AI "refinement" loops, the vocabulary diversity (type-token ratio) collapses.)

Semantic ablation: Why AI writing is boring and dangerous

opinion: The subtractive bias we're ignoring

(www.theregister.com)

[graydon@canada.masto.host]

@cstross I find it hard not to wonder if the attraction of AI doesn't include so much forcible normalization of text that there is never anything unfamiliar, and thus never any reason to question a self-assessment of "smart".

(So much gets so messed up by mistaking responses for properties.)

[oddhack@mstdn.social]

@cstross on my rare exposure to the FB algo. feed, I see a long stream of heart-wrenching pleas for cats needing adoption, all matching a very standard and repetitive structure. I suspect the Cats are real and need help, but the pleas are "AI".

ActBlue (Former US Democratic political donation platform) suffered this even pre-LLM with its endless torrent of nearly identical begs for more money.

[cturnbow@mastodon.world]

@cstross

Reducing the text to the most common denominators?

[agowa338@chaos.social]

@cstross

Or in other words "Why AI is writing in corporate speech"

[pattykimura@beige.party]

@cstross

Really great article, thanks for posting.

I know those who use ChatGPT to improve professional work letters. AI ChatGPT doesn't "improve" writing to refine meaning and nuance, it is designed to dumb down writing in successive loops to the most middling 6th grade level of comprehension thereby "improving" its "reach". It's the specifically designed feature of AI "editing".

I recently read a ChatGPT "improved" letter that may violate the Civil Rights Act by using a term that is a red flag for a lawsuit. I alerted the sender who confessed to using ChatGPT because her boss uses it and raves about it.

An ungraceful letter that keeps you legal is safer than a "clean" "professional" GPT letter that gets you sued. Know the risks!

[kats@chaosfem.tw]

@cstross Blandness as a service.

[sarble@mefi.social]

@cstross This gives some formal insight into the feeling I have when reading AI extrusion, which is that my eyes slide right off it. It's like eating one of the varieties of apple that having been excessively bred for appearance and not taste, the promise of substance so at odds with reality that my brain revolts.

[akamran@indieweb.social]

@cstross that explains why everything ends up sounding like a middle management wannabe on LinkedIn

[cigitalgem@sigmoid.social]

@cstross "beigification' is aspect of the recursive pollution problem

Recursive Pollution and Model Collapse Are Not the Same | BIML

Forever ago in 2020, we identified "looping" as one of the "raw data in the world" risks. See An Architectural Risk Anal

Berryville Institute of Machine Learning (berryvilleiml.com)

[fosstastic@mastodon.social]

@cstross Well, that explains why every time I tried #AI for improving a text snippet, I was very disappointed.

It always wants to convince me that I should remove any uncommon sentence structures and replace them with generic ones, which often removes any personality from written text.

If you ask it to make a small addition to an existing text, it likes to rephrase everything in a more generic way and is unable to add a subtext layer. Honestly, it's just useless for writing in my opinion.

[meuwese@mastodon.social]

@cstross this is indeed a very neat explanation why the best possible outcome of an LLM is still terrible.

[steltenpower@social.edu.nl]

@cstross As someone who, next to having a little knowledge about LLMs, was once complemented for choice of words by a native-speaking professor I hung around with at a conference for a few days, I am not surprised about this LLM fact. The professor however then was somewhat surprised of his own uttering and continued "but I had a few beers".

[numodular@c.im]

@cstross Seaticks skewed row:bots metrics

AI, Captain!

#BoycottAI

[mmcknett@hachyderm.io]

@cstross “semantic ablation” is a concise way to describe that feeling of “I just read all the words but I can’t tell what they are trying to convey” that I have gotten after reading certain generated snippets of text.

[pieist@ohai.social]

@agowa338 @cstross

Depends what level of the corporation you''re speaking to. Management and marketing hype is in some ways the opposite, with a heavy use of "signaling" words that serve little informational purpose, rather they are meant to leave an impression. Think "disruptive", or "leverage" used as a verb...

[cstross@wandering.shop]

@malice @JdeBP The Register is a news site: everything has to be flensed and filed down to fit in a standard format and voice. That piece is probably all that's left of an original that was three times the length.

[subterfugue@sfba.social]

@cstross there’s worse related things.

We come to see anything that the AI can not and does not produce as invalid and thus reading these bullshit, taupe texts shrinks our creative range, our sense of the possible, and our willingness to forge our own path or follow someone else blazing their own.

Narrowing the range of semantics to an average is one thing.

Strangling our range of ideas is another.

[agowa338@chaos.social]

@pieist @cstross

Well but you'll still notice that they're using a bunch of words to say nothing legally binding at all in the end...

[greenviv@social.vivaldi.net]

@cstross It is impossible to replace the human experience with a machine. The moment by its nature is sancrosanct; it's only in this atmosphere of gaming real estate insanity where life's nature is just another bitcoin to earn where we have lost our way.