BS Meter

"BSMeter.online scores YouTube video truthfulness 0-100%. AI + human wisdom analyzes claims and detects misinformation. Creators improve content, users share results, truth goes viral.

Comment

Inspiration

I left my regular six figures nine-to-five so I could pursue a few ideas from a few channels on YouTube. I was not naive insofar as I knew most of the content were blatantly false. But what if I could find one statement, one idea, one strategy that I could capitalize on, the financial freedom would be a dream for my family and I. I am still searching for that dream. But what if there was a tool that could help me verify most of these claims, then and only then, I would succeed in my quest.

Based on my personal story and with over 2 billion of hours and views daily on YouTube, I saw an urgent need to help viewers like myself validate quickly and effectively with the use of AI claims from our favorite channels. The channel owners would in turn benefit from the feedback, increase their credibility and be empowered in their content.

What it does

BSMeter.online analyzes Youtube videos for accuracy using multiple LLMs for consensus and evaluation. Users paste a Youtube URL and within 30 seconds, our application transcribes the video's audio, extracts of the individual claims, and fact-checks the statements using ChatGPT, Claude and Gemini. Once done, the site generates a comprehensive score and a statement by statement score. Because AI is not perfect, users from the sites and community can also vote on the scores and statements. Finally, the owner of the video can claim it and improve their content's authenticity based on the feedback.

How we built it

Using Bolt.new as our primary tool, we were able to produce a front-end that communicates with Open AI, Claude and Gemini for transcription and AI evaluation. For database, we leverage the integration of Bolt.new with Supabase and Netlify for Deployment. We use Ionos for domain. Further, we use Google Auth alongside basic email and password for authentication. Finally, we use Google Youtube API v3 to work with the video details

Challenges we ran into

Most of our challenge was to make sure for a simplify UX experience while providing context and details where necessary. Further, it took us some time to balance on when to call the different AI elements in effectively both from cost and usage perspectives. Qualifying the AI answers also, would be an ongoing feat as we grow the platform.

Accomplishments that we're proud of

That we've created a tool for viewers, creators, the Youtbe platform and the whole community to have one central and common point: clear and unbiased fact-checking content from videos. We are also proud that this was done this in a short period of time with the help of Bolt.new and getting multiple AI models working together.

What we learned

We learned using multiple AI Models increases the credibility of the platform itself and having a community feature was paramount to balance the AI sometimes imperfect dynamics. We wanted to get it right from the jump. We remain confident the intersection of AI, community features and viral mechanics will continue to create stronger effect and better results. Human and AI will continue to work together to improve the World as we know it. Daily, in real-time at all levels.

What's next for BS Meter

We would like to get funded so we can expand beyond Youtube and address other platforms and medium such as Tiktok, podcasts, and live streams. Further, we would like to develop advanced creator dashboards for their channels, build API partnerships with news organizations and educational institutions. Finally, allow enterprise users of the platform to white label it or bring their own Large Language Models

Credential to login and test

email: amesmin@live.com password: Btines1!

Or you can create your own credentials

Built With

Share this project:

Updates