Biasly - Logical Fallacy Checker

Meet Biasly: Your Shield Against Fallacies! 🛡️ Detecting 20+ logical flaws, it empowers you to fight misinformation with clear explanations. Unveil weak arguments and champion critical thinking!

Comment

Inspiration

In an era dominated by information overload, the battle against misinformation is more critical than ever. Misleading articles permeate our digital landscape, driving political polarization and undermining informed discourse. To combat this, we believe in the power of critical thinking. However, the sheer volume of information we encounter daily makes it challenging to remain vigilant against logical fallacies.

What it does

Enter Biasly, your ally in the fight against flawed arguments. Harnessing the capabilities of Cohere's LLMs, Biasly is a revolutionary tool designed to automatically detect over 20 logical fallacies in writing. With just a click, you can unveil poorly constructed arguments and gain valuable insights into their flaws. But we don't stop there - Biasly provides clear explanations for each identified fallacy, empowering users to understand and counter misinformation effectively.

Detecting 24 logical fallacies:

  • Strawman
  • False Cause
  • Slippery Slope
  • Ad Hominem
  • Special Pleading
  • Loaded Question
  • The Gambler Fallacy
  • Bandwagon
  • Black or White
  • Begging the Question
  • Appeal to Authority
  • Composition/Division
  • Appeal to Nature
  • Anecdotal
  • Appeal to Emotion
  • The Fallacy Fallacy
  • Tu Quoque
  • Personal Incredulity
  • Burden of Proof
  • Ambiguity
  • No True Scotsman
  • Genetic
  • The Texas Sharpshooter
  • Middle Ground

How we built it

Biasly leverages advanced AI capabilities in a two-step process. First, it employs Cohere Classify, a powerful tool that scans texts to identify logical fallacies. This step serves as the backbone of Biasly's functionality, allowing it to pinpoint flawed arguments accurately. Once a fallacy is detected, Biasly seamlessly moves to the second step, utilizing Cohere Generate. This feature provides clear and concise explanations for each identified fallacy.

Challenges we ran into

  1. Getting good examples for the Cohere Classify API
  2. Building intuitive and interactive UI

Accomplishments that we're proud of

  1. Built a reasonably accurate classifier for 24 different logical fallacies
  2. Expanded the classifier with generative technologies to provide more guidance to user
  3. Cool UI

What we learned

  1. Understanding LLMs and its potential use case
  2. Knowing Cohere API inside-out
  3. Learning tailwind CSS for the first time and loved it!

What's next for Biasly - Logical Fallacy Checker

  1. Chrome extension to streamline user experience
  2. Multi-sentences classification
  3. Custom model to reduce REST payload to the Cohere API
  4. Minor UI improvement like loading bar and animation

Built With

Share this project:

Updates