Developed through a consultative process across teaching, research, and administration, this AI strategy responds to the growing presence of AI within our community and in the wider academic and policy landscape. As universities worldwide refine their approaches, this is an opportune moment for the Institute to articulate its own clear, coherent, and responsible framework. Anchored in the Institute’s Charter and mission, it provides a structured yet evolving orientation to guide the thoughtful and principled adoption of AI. The following operational principles underpin the Institute's adoption and use of AI:
Human Oversight: AI systems must allow meaningful human oversight, review, and correction. Fully automated decisions affecting learning, research, or institutional processes are not permitted.
Mission-Aligned Use: AI tools may be adopted only when they demonstrably support the Institute’s academic, educational, administrative, or outreach mission. Adoption driven solely by novelty, convenience, or efficiency gains is insufficient.
Use Boundaries and Disclosure: Uses of AI must be explicitly defined and communicated. Significant AI assistance in academic, research, or professional outputs must be disclosed in accordance with institutional guidance.
Secure and Lawful Data Handling: AI tools must comply with data protection, confidentiality, and intellectual property requirements. Preference is given to solutions that minimise data exposure, allow anonymization or pseudonymization, and offer contractual guarantees regarding data use and storage.
Proportionate Cost and Environmental Awareness: Financial, organisational, and environmental costs must be assessed prior to adoption. Solutions should be scalable, sustainable over time, and proportionate to the benefits they deliver.
Transparency and Reproducibility: AI outputs and recommendations must be explainable, reviewable, reproducible, and open to challenge. AI applications and systems should allow scrutiny of their assumptions, data sources, and limitations, and enable the documentation and replication of AI-assisted processes where relevant.
Equitable Access and Usability: AI tools should be accessible to users with diverse levels of technical expertise and support multilingual use where possible. Adoption decisions must consider impacts on equity, accessibility, and inclusion.
Continuous Evaluation and Adaptation: AI tools and practices are subject to ongoing monitoring, periodic review, and revision. Tools may be modified, restricted, or discontinued in light of new risks, increase in licensing costs, or institutional priorities.
AI Literacy: AI adoption must be accompanied by sustained investment in AI literacy, including guidance, training, and institutional support, to ensure informed, critical, and responsible use. Responsibility for ethical practice and risk management cannot rest solely on individual users.
-----
The consultative process was conducted over the course of 2025 and jointly coordinated by the Tech Hub, the Direction of Studies, the Research Office, the Direction of Administration, and the GISA Student Association, under the supervision of the Direction of the Institute.
Further details on the Institute's AI Strategy and accompanying action plan will be shared in the coming weeks. For more information, please contact the Tech Hub: techhub@graduateinstitute.ch
