Original briefings, reports, and essays from across Alesvia's initiatives. Evidence-led perspectives on how AI shapes human autonomy, institutions, and public life.
Recent interpretability research suggests that internal emotion-like representations can affect model behavior under pressure. That should change how institutions evaluate and govern AI systems.
AI is no longer only a technical tool. It is becoming part of the environment in which people think, decide, and seek guidance. That is why autonomy, not generic ethics language, has to become the governing standard.
Most institutions do not suffer from a lack of principles. They suffer from a lack of standards, implementation logic, and operational guidance. The next phase of serious AI governance has to move there.
Most teams evaluate conversational AI for speed, cost, and accuracy. Far fewer audit it for dependency, escalation failure, autonomy erosion, misleading emotional fluency, and pressure behavior. That is a serious blind spot.
The EU AI Act mandates AI literacy under Article 4, but most organizations have no idea what that means in practice. Here's our framework for getting it right.
A Brown University study found AI chatbots systematically violate mental health ethics. Yet most therapists have never used one. We need to close this gap — fast.
70+ AI companies, $248M raised, and a government investing billions. Here's our first comprehensive mapping of the Czech AI landscape — and where human autonomy fits in.