Position of the ifo Institute, CESifo, and EconPol on the Use of AI
How do the ifo Institute, CESifo, and EconPol use artificial intelligence (AI) in research, administration, and communications? Our shared position outlines its uses and the principles of transparency, human review, and responsibility.
AI is not only a subject of our research but also a tool for employees in all areas – research, administration, and communications. The ifo Institute expects its employees to use AI in their daily work, and we are creating an environment that encourages innovation to support this. At the same time, we are aware of the risks associated with this tool, including:
- Misinformation from unreliable sources
- Mainstreaming and homogenization of content
- Reproduction of biases
- Inefficient use of resources
- Risks to intellectual property and data protection
Scientific, editorial, and legal responsibility for our work lies entirely with the authors. We are transparent about the use of AI in producing publications; the following sections set out how. The ifo Institute does not publish scientific work generated purely by AI.
Research
In our research, we make use of the opportunities AI offers, for example in coding, data analysis, and the structuring of large datasets or the detection of patterns within them. In scientific publications, we document the system used, the purpose, and the extent of its use, where required by good scientific practice. Our content continues to rest on our researchers’ independent development of research questions, their choice of suitable methods, their verification of sources, and their expert assessment. The use of AI does not change the standards for scientific work set out in the ifo Institute’s Code of Conduct: Researchers bear responsibility for the creation, publication, and communication of research content and findings.
Management & Services
In Management & Services, we use artificial intelligence to make our work more efficient and improve its quality. This includes analyzing aggregated data for financial and human resources management and preparing complex decisions.
In Communications, we use AI to create and distribute content more efficiently, for example for research support, translation, and text editing, as well as for creating graphics and editing image, audio, and video material. We label AI-generated or substantially AI-edited content in accordance with legal requirements and beyond, where this matters for how our audience interprets it.
We process confidential and personal data only in anonymized form or in approved systems, and in compliance with data protection requirements. We review AI-assisted results in particular for factual accuracy, traceable sources, possible bias, and copyright and data protection risks. Decisions are always made by people.
Notes for Publication Imprints
- Text: ifo texts may be created in part with the support of generative AI, but they undergo editorial and subject-matter review by humans before publication.
- Text assistance: Generative AI is generally used for language editing and translation of ifo Institute texts.
- Images: Photos in the ifo Institute’s media are never generated purely by AI, but their quality is enhanced with AI tools.
- Graphics: Graphics at the ifo Institute are created in part with the help of generative AI. The data basis and editorial responsibility lie with the author.
- Video/audio: Real people appear in the ifo Institute’s audiovisual communication formats. The ifo Institute uses AI tools for editing and for generating animations.