Saturday, 28 June 2025

I asked Microsoft's Copilot AI an obvious question...

 What are the tell tale signs of AI writing?

There are definitely some fingerprints that AI tends to leave behind—especially when it's not trying to hide. Here are a few of the most common tells researchers and editors have spotted:

  • Overuse of formal or Latinate vocabulary: Words like “utilize,” “commence,” or “ameliorate” pop up more often than they would in natural human writing, especially when simpler words would do.

  • Repetitive phrasing or tautologies: AI sometimes says the same thing twice in slightly different ways—e.g., “Let’s come together to unite” or “a true fact.”

  • Paragraphs that start with discourse markers: Words like “However,” “Moreover,” or “Overall,” followed by a comma, are a common AI tic.

  • Frequent use of the Oxford comma: That’s the comma before “and” in a list—AI tends to use it consistently.

  • Numbered lists with colons: AI loves structure, and it often defaults to tidy, formulaic formats.

  • Bland or journalistic tone: The writing can feel polished but emotionally flat—like it’s trying to sound neutral or encyclopedic.

  • Overuse of em dashes: AI often leans on these to create rhythm or insert asides—sometimes excessively.

  • Lack of deep analysis or personal insight: Even when the content is informative, it may lack the nuance, contradiction, or vulnerability that marks human thought.

That said, these signs aren’t foolproof. A skilled human can mimic AI, and a well-tuned AI can mimic a human. But when you start seeing several of these traits together, it’s a strong clue.

Would you like to test a passage and see if it raises any red flags? Or maybe explore how to blend your own voice with AI tools without losing that human spark?

No comments:

WHAT WAS IT? AND OTHERS by Fitz-James O'Brien

The final volume of Collected Speculative Works from Swan River Press covers the period 1858 to 1864. It was in the opening stages of the Am...