ai_test_guiding

Guiding Principles and Ethical Commitments

The University of Illinois System is committed to ethical, transparent, and responsible AI use.

Human Oversight and Autonomy

You are responsible for all AI outputs you use or share. Verify accuracy and maintain human review of consequential decisions. AI should support, not replace, human judgment, especially in academic, clinical, and administrative decision-making. Establish clear accountability and remediation procedures for harm or bias.

Transparency and Honesty

Disclose AI use when appropriate, especially in academic and clinical settings. When AI meaningfully contributes to academic, clinical, research, or administrative work, that role should be disclosed to maintain trust and academic integrity. Communicate the technology's capabilities, limitations, and potential biases to users and stakeholders. To the extent practical and appropriate provide transparency in algorithms and decision-making processes to enhance trust and accountability.

Privacy, Security, and Data Stewardship

Protect high-risk data. Never input sensitive, confidential, regulated, or proprietary information into non-approved AI tools. Protecting privacy and minimizing unnecessary data exposure is a core ethical obligation. Ensure AI systems incorporate privacy and security by design, comply with applicable laws, and implement strong privacy and security measures to prevent unauthorized access. Data used to train, refine, or prompt AI must be handled in alignment with university policies and legal requirements.

Fairness and Inclusion

Recognize AI can reflect biases from training data. Critically evaluate outputs, especially for decisions affecting people. AI must not be used in ways that create or reinforce discriminatory outcomes. Use inclusive design practices, identify and mitigate biases, and seek input from diverse stakeholders to prevent discrimination and address potential barriers or unintended consequences. Users should actively evaluate outputs for potential bias and engage diverse perspectives when deploying or interpreting AI systems.

Monitoring and Awareness of Impact

AI systems can degrade or drift over time. Regularly reassess deployed AI tools to ensure continued performance and accuracy. Initial validation does not guarantee ongoing reliability; periodic review is essential to maintain system effectiveness and identify emerging issues. AI tools can shape behaviors, decisions, and outcomes. Users should consider downstream impacts, including unintended consequences, and avoid applications that could cause harm or diminish equity, trust, or safety.

Continuous Learning and Reflection

AI technologies evolve rapidly and best practices continue to mature. Establish feedback mechanisms to capture user insights on system performance, emerging risks, and unintended consequences. Integrate feedback into ongoing governance processes to adapt policies and practices as the technology and our understanding advance. Users are expected to stay informed, reflect critically on their use of AI, and raise concerns when potential ethical issues arise.