11 comments

[ 0.27 ms ] story [ 23.2 ms ] thread
I love the term "unslop" when of course that's the exact opposite of what you're doing. Very evil. Much wow.
The quiz is easy, humans use smaller words because they are not as well read as AI.

However I couldn't do the quiz because after one or two questions I was forgetting what I was supposed to be clicking, human text or AI text.

The UI should visually confirm what you just makred and as what and have a button to proceed to next question when you are satsified with your selection.

I don’t think this is working that well. The unslopped content has a whiff of automatic translation. In the same way that it is impossible to get an LLM to sound authentically British.
Reminds me of the whole "no-makeup" makeup concept.
It's as simple as adding to the prompt beforehand or asking for the style to be modified afterward, but the hallmark of most slop-mongers is that they don't actually care enough to put in even that much effort.
(comment deleted)
But I already have an effective method of removing AI voice from writing.
One trick is to ask the AI to reword something for an N-th grade reading comprehension level for some N <= 8.

Not that I do this (I don't write with AI), but I have experimented with it.

Here is what Gemini 3.6 Flash produced from the example paragraph when I requested a rewrite to 6th grade reading comp:

Growing a newsletter audience takes time and a good plan. You need to send your newsletter regularly and figure out what topics your readers like best. When you share things they care about, they will keep reading. Remember, building a big audience doesn't happen overnight—it takes steady work, but it pays off in the end!

There is still the snappy take-away pontification at the end with an emdash, but it no longer reads like corporate speak. It avoided hackneyed crap terms like: multifaceted strategy, delve into, resonates, leverage, insights, foster, engagement, sprint not marathon, embracing, sustainable, mindset.

100% for me on the quiz. Robots are great, but they use overly complex constructions at the wrong time.