In today's digital age, an astonishing 70% of the information circulating online contains some form of bias or misinformation, shaping perceptions and influencing decisions unknowingly. This alarming statistic underscores the critical need for media literacy in our increasingly interconnected world.
That's where we come in! Misinfo is an innovative web app designed to hone your ability to spot misinformation. Leveraging the power of Intel's cutting-edge language models, we offer users a unique platform to explore topics of interest. Dive into a diverse array of online content, from articles to news extracts. But here's where it gets intriguing: amidst this content, we've used the power of generative AI to interlace some articles with one of six distinct types of bias. The question is, can YOU tell the difference?
Building Misinfo was an exciting journey that combined cutting-edge technologies and creative problem-solving. Here’s how we brought it to life:
- A Modern and sleek front end and UI, built using React, with Tailwind CSS for styles
- We feature a robust and fleshed out the backend, built fully in Python with Flask
- We utilize a web-scrapper to build a set of rich, diverse range of online excerpts from articles
- We use Intel's powerful AI capabilities to train an LLM (GPT) that learns to smartly add bias
- We report a dashboard to the user that informs them about how many articles they could guess!
However, innovation is a journey, and we'd like to take a moment to share some areas where we thought outside the box and others where we hope to expand our product!
Initially, our AI model was far too inefficient, taking upwards of 20 seconds to process each data point, so we creatively built an algorithm that allows the model to cache multiple generated articles at a time so that upon query, we'd have a way of avoiding re-prompting
While we initially created a results dashboard that shows analytics on the user's reaction to the articles, we hope that with more time and resources, we can expand these analytics to a suite of metrics that add additional value!
With media bias and the spread of misinformation running rampant, we're proud to create a unique full-stack app that raises awareness by actively showing netizens how easy it is to fall for misinformation. We hope that users can take away an experience that reminds them how important it is to make the effort to find factual, diverse information. We specifically designed our interface drawing inspiration from high-activity apps to appeal to the younger generation of netizens, who comprise most of the demographic. We're also proud of our innovative application of generative AI to solve a unique, but pressing problem in our community!
In the process of combating misinformation, we learned a lot! Delving deeper into full-stack development, each team member enhanced their expertise in both front-end and back-end technologies, crafting a cohesive and robust platform. Our exploration into ML/AI methods was particularly enlightening, pushing the boundaries of what we thought possible with Intel's Large Language Models (LLMs). Through creative problem-solving, we refined our AI algorithms, achieving a delicate balance between complexity and performance.
Working as a close-knit team, we uncovered the undeniable value of community. It became clear that Misinfo's potential was not just in educating individuals but in fostering a community that values truth and critical thinking. This realization has profoundly shaped our vision for the future.
Our next steps involve hosting the platform and actively sharing the link across social media and other popular digital spaces. Our goal is to ignite conversations about misinformation, encouraging a collective effort towards a more informed society. We believe in the power of community to drive change, and by making Misinfo accessible to a wider audience, we hope to inspire a grassroots movement against misinformation.
Log in or sign up for Devpost to join the conversation.