AIcitizen

Critical Thinking AI - Combating Misinformation Through Education and simulations.

Inspiration

Seven years ago, I created a project focusing on logical fallacies, to help people recognize faulty reasoning and manipulative arguments in everyday speech. This project opened my eyes to the vulnerability of each and every one of us to sophisticated manipulation, particularly in the digital age. Faced with the explosion of AI-generated content, deepfakes and increasingly sophisticated disinformation campaigns, I realized that my work in detecting sophistry was just the beginning. The world desperately needs tools and education to help people develop true critical thinking, not only to spot the obvious lies, but also to navigate the gray areas where truth and falsehood mix. The inspiration crystallized when I noticed that even intelligent, educated people had trouble distinguishing between AI-generated content and authentic human expression, between cherry-picked statistics and meaningful data, between correlation and causation in news articles.

What it does

I create plausible scenarios for personal training to sharpen analytical thinking, based on the fundamentals of recognizing logical errors to build comprehensive media literacy and critical thinking skills.The platform presents users with real-life scenarios, news articles, social media posts, political speeches, scientific claims, and marketing messages. It then guides them through systematic analysis techniques. Unlike simple fact-checking tools, it teaches users to assess sources, recognize emotional manipulation, spot misleading statistical representations, and identify missing context or cherry-picked evidence.

The AI does not simply tell users what is true or false; it supports them through the reasoning process, showing how to break down complex claims, cross-check sources, and maintain healthy skepticism without falling into cynical nihilism.

The system develops critical thinking through progressive challenges, starting with obvious examples of manipulation and gradually introducing more sophisticated forms of misinformation and bias. Users learn to recognize not only outright lies, but also truthful statements presented in misleading ways, exaggerated legitimate concerns, and genuine expertise mixed with unfounded speculation.

How I am building it

Building on my seven-year experience with the logical fallacies project, I had a deep understanding of how people actually learn to recognize manipulative reasoning. It's not about memorizing lists of fallacy names, but about developing intuitive pattern recognition and systematic questioning habits. We created an AI system trained on vast datasets of verified true and false information, but more importantly, on the reasoning processes that experts use to evaluate claims. The technical architecture combines natural language processing for content analysis, machine learning models for bias detection, and interactive dialogue systems that can engage users in Socratic questioning.

The platform integrates real-time content from news sources, social media, and academic publications, automatically identifying teachable moments where users can practice their skills on current events. We developed sophisticated algorithms that can detect subtle forms of manipulation like selective quotation, false equivalencies, and appeals to emotion disguised as logical arguments. The system maintains detailed profiles of each user's critical thinking development, identifying their particular vulnerabilities and blind spots, then designing personalized training sequences to address those weaknesses.

Challenges I ran into

I am still in development and am primarily seeking beta testers for the scenarios, as well as greater visibility for this common-sense project. I am also working on the educational component in collaboration with university professors.

The greatest challenge was avoiding the trap of creating another echo chamber or ideological filter. Critical thinking isn’t about teaching people what to think, but how to think, and we had to be extraordinarily careful not to embed our own biases into the system. We discovered that people’s existing beliefs strongly influence what they consider “critical thinking” conservatives and progressives often have completely different ideas about which sources are trustworthy and which analytical frameworks are legitimate. Building an AI that could teach genuinely neutral analytical skills while respecting diverse worldviews required constant testing with users across the political spectrum.

Technical challenges centered around the subtlety of human reasoning and the context-dependent nature of truth evaluation. Teaching an AI to recognize when a statement is technically accurate but misleading in context, or when statistical data is presented in ways that distort its meaning, required developing new approaches to semantic analysis and logical reasoning. We also struggled with the emotional dimension of critical thinking people often reject valid information that conflicts with their beliefs, not because they lack analytical skills, but because accepting the information would be psychologically uncomfortable.

Accomplishments that I am proud of

As for progress, I will truly be satisfied when I see people becoming more aware, taking responsibility for their own understanding, and actively shaping the future. My goal is to empower individuals to think independently and critically, so that they are equipped to make informed decisions and contribute meaningfully to society.

What I learned

The most important insight I gained is that critical thinking is less about intelligence or education and more about intellectual humility and systematic habits of mind. Many highly educated people struggle with critical thinking because they are overconfident in their knowledge and resistant to change. On the other hand, some individuals with little formal education excel at critical thinking because they approach information with healthy skepticism and methodical questioning. Effective critical thinking education must address not only cognitive skills, but also the emotional and psychological barriers that prevent clear reasoning.

I also discovered that the misinformation landscape is far more sophisticated than most people realize. State actors, commercial interests, and ideological groups use increasingly subtle manipulation techniques that exploit cognitive biases most people are unaware of. This experience reinforced my conviction that critical thinking education is not a luxury, but a democratic necessity in the digital age.

What's next for AIcitizen

The next phase of the project is to recruit alpha testers who can provide valuable feedback on the platform’s scenarios and overall user experience. I will then focus on expanding communication and outreach efforts to increase visibility and attract a broader audience. Although I am currently developing the project in my spare time, my long-term objective is to dedicate myself to it fully and build a comprehensive platform featuring engaging, gamified content.

Initial testing with three scenarios has shown promising results: users spent an average of 90 minutes on the site, were deeply engaged in logical reasoning, and offered positive feedback. This confirms the project’s potential and motivates me to continue refining both the scenarios and the educational modules. As the platform grows, I am actively seeking technical support, particularly from an experienced developer, to address emerging challenges and ensure sustainable progress.

Built With

• autoprefixer
• chrome
• css3
• edge;-cloud-services:-supabase-cloud
• fetch-api;-other-technologies:-eslint
• firefox
• framer-motion
• html5
• javascript
• json
• languages:-typescript
• local-storage
• node.js
• postcss
• progressive-web-app-features
• react-router-dom;-platforms:-supabase
• responsive
• safari
• session-storage;-apis:-supabase-rest-api
• sql;-frameworks:-react-18
• supabase-auth
• supabase-auth-api
• supabase-database
• supabase-realtime;-databases:-postgresql
• supabase-storage
• tailwind-css
• uuid
• vite
• web-speech-api
• websockets