About this video

• Video Title: I Can Spot AI Writing Instantly — Here’s How You Can Too
• Channel: Evan Edinger
• Speakers: Evan Edinger
• Duration: 15:57

Overview

This video teaches viewers how to identify AI-generated text by highlighting common red flags. Evan Edinger explains various indicators such as the overuse of M-dashes, parallel sentence structures, convenient lists of three, uncanny word choices, and exaggerated praise, providing examples from emails, social media posts, and even political statements. He contrasts these with characteristics of human writing, like personal anecdotes, tangents, and genuine emotion, to help viewers distinguish between the two.

Key takeaways

• Overuse of M-dashes: AI models tend to use M-dashes more frequently than humans, making it a noticeable indicator.
• Parallel sentence structures: AI often employs repetitive sentence patterns like "It's not just X, it's Y," which can sound unnatural.
• Convenient lists of three: AI frequently uses triplets (groups of three) in lists, likely due to human preference for this structure in writing and visual composition.
• Uncanny word choice and tone: AI-generated text can sometimes feel "off" due to unusual word usage or a strange tone, often described as the "uncanny valley of the English language."
• Exaggerated and empty praise: AI responses may include overly positive or generic compliments that lack genuine sincerity.
• Unusual analogies and similes: AI can generate analogies that are technically correct but feel forced or nonsensical, like "a band-aid made of sandpaper."
• Constant restatement and over-explanation: AI writing often repeats points unnecessarily or provides excessive preamble, making it less concise.
• Generic content and lack of personal touch: Human writing typically includes personal experiences, anecdotes, and a unique voice, which are often missing in AI-generated text.
• Perfect grammar in informal contexts: While not a definitive sign, consistently perfect grammar in informal online communication (like iMessage) might suggest AI assistance.
• Tangents and narrative journeys: Human writing often includes relevant tangents and non-linear narratives, making the content more engaging and less direct than typical AI output.

The three red flags identified in the LinkedIn post about photography are:

1. Use of M-dashes: The sentence included an M-dash, which AI models frequently use.
2. Parallel sentence structure: The sentence followed a pattern like "It's not just X, it's Y."
3. Convenient list of three: The sentence contained a list of three items: "telling a story, conveying emotion, and creating a visually compelling image."