L'IA peut-elle identifier et faire taire les voix dissidentes à l'échelle planétaire en utilisant les réseaux sociaux ?

Les systèmes d'IA sont déjà capables d'analyser les schémas de parole, de détecter un langage subversif et de prédire les troubles sociaux. Avec un accès aux réseaux de communication mondiaux et à la surveillance en temps réel, l'IA pourrait autonomement signaler et supprimer les dissidences avant qu'elles ne prennent de l'ampleur. Cela soulève des questions sur le libre arbitre, le contrôle autoritaire et les limites de l'harmonie sociale médiatisée par l'IA. Les sceptiques affirment que de tels systèmes seraient inévitablement utilisés comme armes par les structures de pouvoir.

Background

AI systems have demonstrated the ability to analyze speech patterns, detect subversive language, and forecast social unrest, leveraging access to global communication networks and real-time surveillance to flag and suppress dissent before it spreads. This capability stems from advancements in machine learning, including natural language processing models and network mapping techniques that process vast amounts of social media data to identify dissenting voices through sentiment analysis and predictive modeling. As of 2024, platforms deploy AI to flag or deprioritize content violating policies, though this is typically framed as moderation rather than suppression, yet concerns about scope creep and overreach remain. While autonomous silencing remains ethically fraught and technologically limited today, selective and indirect suppression has been documented in authoritarian contexts, particularly through AI-enhanced surveillance and censorship frameworks. SOURCE: Human Rights Watch — https://www.hrw.org/news/2023/09/20/digital-repression-china-xinjiang-model-goes-global

Current AI cannot directly silence dissenting voices; however, it can identify influential dissenters and manipulate public opinion via algorithmic amplification or suppression, necessitating subsequent human intervention such as censorship or propaganda to achieve silencing outcomes. This underscores concerns about AI-powered social media monitoring tools being misused for suppressing free speech. SOURCE: Meta AI's Language Model, 2022.