Glossario IA
Il dizionario completo dell'Intelligenza Artificiale
AI Governance
Set of processes, policies, and organizational structures ensuring the responsible development, deployment, and maintenance of AI systems compliant with regulations and ethical values.
Model Card
Standardized technical document providing information about the characteristics, performance, limitations, and intended use of a machine learning model to ensure its transparency.
Datasheet for Datasets
Detailed documentation describing the provenance, composition, potential biases, and usage recommendations of a dataset used to train AI models.
Explainable AI (XAI)
Set of techniques and methods allowing for understanding and interpreting the decisions of AI models, essential for auditability and stakeholder trust.
Drift Detection
Process of identifying changes in the distribution of input data or the relationship between features and target that can degrade model performance.
Ethical AI Framework
Methodological structure defining the principles, guidelines, and controls to ensure the ethical development and deployment of AI systems.
Algorithmic Auditing
Systematic and independent examination of algorithms to assess their compliance with regulatory, ethical, and performance requirements in production.
Bias Mitigation Techniques
Methods applied to data, models, or predictions to reduce or eliminate systemic biases that can lead to discrimination in AI systems.
Responsible AI
Holistic approach integrating ethics, transparency, fairness, and responsibility throughout the entire lifecycle of AI systems to minimize societal risks.
ML Lifecycle Management
Structured management of all machine learning phases, from data preparation to model retirement, with complete traceability and documentation.
Model Interpretability
Ability to understand and explain a model's internal workings and the reasons for its specific predictions, crucial for auditability.
Compliance Automation
Use of automated tools to check, document, and ensure continuous compliance of AI systems with applicable regulations and standards.
Ethical Risk Assessment
Systematic evaluation of potential ethical risks associated with deploying an AI system, including social impacts, discrimination, and privacy violations.
Transparency Reporting
Regular documentation of AI systems' practices, performance, and impacts to maintain stakeholder trust and meet regulatory requirements.
Model Documentation
Complete and structured record of all information related to an AI model, from its design to deployment, to ensure its traceability and auditability.
Regulatory Compliance
Compliance with AI-specific laws, regulations, and standards such as GDPR, AI Act, or sectoral ones in the deployment and operation of intelligent systems.
Stakeholder Impact Assessment
Systematic analysis of the potential effects of an AI system on all stakeholders to identify and mitigate negative impacts before deployment.
AI Ethics Board
Multidisciplinary committee responsible for overseeing ethical issues in AI development and deployment, ensuring alignment with organizational values.