HOW AI FEED LENS
ACTUALLY WORKS
A transparent breakdown of how our extension identifies synthetic writing patterns, where processing occurs, and how to interpret probabilistic classifications.
The 5-Step Detection Lifecycle
Install the Browser Extension
Add AI Feed Lens to Google Chrome or Microsoft Edge with a single click. The extension requires zero invasive permissions—no access to your private messages, contacts, or login credentials.
Navigate to Your LinkedIn Feed
Open linkedin.com as you normally do. AI Feed Lens runs seamlessly in the background without modifying your normal browsing experience or slowing down page rendering.
In-Feed DOM Detection
As posts scroll into view, our lightweight content script identifies standard public post containers. It extracts only the visible text of the post for lexical evaluation.
Linguistic & Statistical Analysis
The post text is evaluated across key linguistic dimensions: perplexity variance, burstiness (sentence length diversity), vocabulary entropy, and formulaic AI transition patterns.
In-Feed Classification Pill Appears
A neat classification badge renders directly inside the post header: HUMAN WRITTEN (purple badge) or AI WRITTEN (red badge), complete with a confidence rating.
✓ What Gets Analyzed
- Public post body text visible in your feed
- Sentence structure, length variations, and vocabulary entropy
- Common AI rhetoric tropes (e.g. "In today's fast-paced world")
- Burstiness and perplexity distributions across paragraphs
✕ What We NEVER Touch
- Your LinkedIn login, password, or session cookies
- Your private messaging inbox or connection requests
- Your personal profile views or search history
- Keystrokes, drafts, or internal browser tabs
Understanding the Classification Labels
Indicates high lexical variance, irregular rhythm, authentic colloquialisms, and idiosyncratic sentence structures characteristic of natural human composition.
Indicates uniform token probability, predictable listicle framing, recycled transitions, and homogeneous paragraph pacing typical of LLM-generated drafts.
Honest Limitations: Why AI Detection Is Probabilistic
We do not make inflated claims of "100% mathematical certainty." Writing detection is inherently statistical:
- Heavily Edited AI Drafts: If a human substantially rewrites an AI draft with personal stories, the metrics naturally shift toward human variance.
- Rigid Human Writing: Highly standardized corporate announcements or formulaic press releases can occasionally exhibit low variance and trigger false positives.
- Informational Tool: AI Feed Lens is designed as an analytical aid for critical readers, not as disciplinary proof or definitive legal evidence.
Ready to install AI Feed Lens?
Experience in-feed post analysis on Chrome and Edge. Free trial included.