Research Focus

Critical Applications

Safe, verifiable and responsible AI deployment in healthcare, justice, finance and environment

Why are these domains critical?

In certain sectors, AI errors are not merely annoying—they are potentially fatal. A wrong diagnosis affects lives. An unfair legal decision violates rights. Undetected fraud causes massive economic harm.

Our research ensures that AI deployed in these domains is verifiable, explainable, fair and resilient. We work with stakeholders from healthcare, justice, finance and environment to ensure that AI adds value while respecting fundamental human values.

Impact Sectors

Critical domains where we work

Salud

Medicine and Diagnosis

Diagnostic support systems that improve accuracy but do not replace medical judgment. Explainability for clinicians and patients.

Justicia

Justice and Security

Recidivism risk systems and evidence analysis that are auditable and bias-free. Transparency for judges and defense.

Finanzas

Credit and Risk

Fair and explainable credit risk assessment. Regulatory compliance and consumer rights protection.

Ambiental

Sustainability

Environmental monitoring, climate change and biodiversity with trustworthy AI. Transparent models for policy-makers.

Trustworthiness Requirements

Attributes that AI must have in critical domains

Rigorous Validation

Extensive clinical/legal trials, validation across diverse populations, continuous post-deployment evaluation.

Security and Privacy

Protection against attacks, privacy guarantees for sensitive data, periodic security audit.

Human Intervention

Meaningful human supervision in critical decisions, effective recourse, preservation of autonomy.

Deployment Challenges

Real obstacles in trustworthy AI implementation

Incompatibility with Existing Workflows

Integrating trustworthy AI requires changing established processes. Organizational friction and resistance to change.

Evolving Regulation

Legal frameworks are still being defined. Uncertainty about future requirements makes proactive compliance difficult.

Cost of Trustworthiness

Validation, audit and transparency have a cost. Balance with pressures of speed and profitability.

Our Research

In the Critical Applications line of the Trustworthy AI group at IAFER, we are developing methodologies to deploy verifiable, auditable and secure AI systems in contexts where the impact is significant.

🚀 Active Research Lines

Multi-scale rigorous validation: Trials and validation in medical, legal and scientific domains
Integration of human supervision: Design of interfaces and processes where humans intervene significantly
Audit and system certification: Frameworks to assess compliance with AI regulations
Post-deployment continuity: Continuous monitoring of performance, security and fairness in production

📚 Publicaciones

Loading publications...