Yapay Zeka Karar Doğrulama Sistemi

warningProblem

"India's top court angry after junior judge cites fake AI-generated orders"

psychologyPotansiyel Çözüm

Mahkeme kararlarının ve resmi belgelerin yapay zeka tarafından üretilip üretilmediğini tespit eden ve doğruluğunu teyit eden bir sistem.

groupHedef Kitle

Hukuk profesyonelleri (avukatlar, hakimler, savcılar), adalet bakanlıkları, üniversitelerin hukuk fakülteleri ve yapay zeka etiği ile ilgilenen araştırmacılar.

paymentsGelir Modeli

Abonelik tabanlı (kullanım sıklığına ve özelliklere göre farklı paketler), kurumsal lisanslama ve API erişimi.

Aksiyon Planı

Metin tabanlı belge yükleme ve analiz özelliği.

Yapay zeka tarafından üretilmiş olabilecek metin kalıplarını ve tutarsızlıkları tespit etme.

Güvenilir kaynaklara dayalı doğrulama modülü.

Sonuçların detaylı raporlanması.

Kullanıcı dostu arayüz.

Pazar Analizi

8.7Puan

Kaynak: Hacker Newsopen_in_new

AI Prompt

Create a single-page Single Page Application (SPA) using HTML, CSS, and JavaScript. The SPA should function as an AI-generated text detector for legal documents. It needs to: 

1.  **User Interface:** 
    *   A prominent title: 'AI Legal Document Authenticator'. 
    *   A large, multi-line text area where users can paste or type text from a legal document. 
    *   A button labeled 'Analyze Document'. 
    *   A clear, designated area below the button to display the analysis results (e.g., 'Confidence Score: XX%', 'Potential AI Generation Indicators: [List]', 'Authenticity Status: Verified/Unverified/Inconclusive'). 
    *   A placeholder for uploading a document file (though for MVP, direct text input is sufficient, but UI should accommodate this for future). 
    *   A simple, professional, and clean design. Use a neutral color palette (e.g., blues, grays, whites).

2.  **Functionality (Client-side simulation for MVP):** 
    *   When the 'Analyze Document' button is clicked, simulate an analysis process. 
    *   For this MVP, instead of complex AI models, implement a *simulated* detection logic. This logic should randomly generate a 'Confidence Score' between 10% and 95% (e.g., 45.7%, 88.2%, 23.9%). 
    *   Based on the generated score, provide a 'Potential AI Generation Indicators' list. These indicators should be pre-defined and appear more frequently for lower confidence scores. Examples: 'Unusual phrasing complexity', 'Lack of specific legal jargon context', 'Repetitive sentence structures', 'Overly formal or generic tone'. 
    *   Determine 'Authenticity Status': 
        *   If score > 75%, 'Verified'. 
        *   If score < 40%, 'Unverified'. 
        *   Otherwise, 'Inconclusive'. 
    *   The results should be displayed dynamically in the designated area.

3.  **Technical Requirements:** 
    *   Use standard HTML5 for structure. 
    *   Use CSS for styling. No external CSS frameworks like Bootstrap or Tailwind are allowed for this specific task. Write custom CSS. 
    *   Use vanilla JavaScript for all interactivity and simulated logic. No external JavaScript libraries (like jQuery) are allowed. 
    *   The entire application must be self-contained within a single HTML file.

4.  **Focus:** The primary goal is to demonstrate the *concept* of such a tool through a functional front-end prototype. The 'AI detection' part is simulated for this MVP.