Introduction
Artificial intelligence research has undergone a period of rapid transformation in both scope and ambition, shaped by advances in machine learning, large language models, autonomous systems and foundational computational theory. At the forefront of these developments in Europe is the ETH Zurich Artificial Intelligence Center, established as a central hub for AI research at one of the world’s leading technical universities. The Centre does not merely aggregate existing expertise; rather, it is designed to foster innovative intersections between fundamental theory, practical implementation and society-oriented applications across disciplines.
The mission of the ETH AI Center is to lead the way towards trustworthy, accessible and inclusive AI systems for societal benefit, grounded in principles of research excellence, people development and impact orientation. To achieve this mission, the Centre adopts an interdisciplinary research model that engages scholars across computational, mathematical, engineering and human-centric disciplines. This paper explicates this model and assesses how the Centre’s research portfolio advances state-of-the-art AI knowledge and practice.
Institutional Model and National Role
The ETH AI Center is founded on the premise that AI research cannot be effectively advanced within disciplinary silos. Rather than operating as an isolated institute, it functions as a central hub across all sixteen departments of ETH Zurich, linking 35 core faculty members with over 100 associate faculty and external collaborators. Its research community encompasses expertise from computer science, electrical engineering, mathematics, cognitive science, robotics, natural language processing, visual computing and more.
Institutionally, the ETH AI Center also plays a key role in national AI initiatives. In partnership with the École Polytechnique Fédérale de Lausanne (EPFL), it co-leads the Swiss National AI Institute (SNAI) and the Swiss AI Initiative, aimed at positioning Switzerland as a global leader in open, transparent and trustworthy AI research. These initiatives underscore a strategic commitment to developing large-scale foundation models and supporting national AI infrastructure, including the deployment of the Alps supercomputer with over 10,000 GPUs to underpin large-scale model training and computational research.
The Centre’s vision explicitly emphasises the development of AI systems that are trustworthy, inclusive and beneficial to wider society, rather than the pursuit of technological dominance alone. This orientation positions the Centre within contemporary debates on ethical AI research and reflects a commitment to responsible innovation.
Interdisciplinary Research Framework
A defining feature of the ETH AI Center is its commitment to interdisciplinary research, which it operationalises across ten core areas encompassing both foundational and application-oriented work. These areas are designed to co-evolve, ensuring that methodological innovation informs practical utility and vice versa.
Foundational AI Research
At the heart of the Centre’s scientific agenda is research that seeks to advance the theoretical underpinnings of artificial intelligence. This work includes:
• Mathematical and statistical foundations of machine learning;
• Safety, reliability and robustness, including the development of models that behave predictably under uncertainty;
• Interpretability and explainability, essential for trustworthy AI systems;
• Fairness and bias mitigation, addressing societal equity concerns;
• Natural language understanding and general cognitive modelling;
• Reinforcement learning and control systems integrated with data-driven insights;
• Privacy-preserving AI, protecting sensitive information in machine learning contexts;
• AI system design and engineering, spanning edge computing to cloud platforms.
This breadth of foundational work reflects a dual emphasis on mathematical rigour and practical impact, recognising that durable AI systems require both robust theoretical frameworks and contextual adaptability.
Application Domains
In addition to disciplinary research in machine learning and computation, the ETH AI Center engages deeply with diverse application areas. These include:
• Robotics and autonomous systems, where perception, planning and control intersect;
• Natural language processing and multilingual AI;
• Human-AI collaboration, focusing on interfaces and augmentative intelligence;
• AI in healthcare and life sciences, encompassing diagnostics, computational biology and biomedical data analysis;
• Economic and organisational AI value creation, exemplified by the Centre for AI Value, which investigates economic potential and AI applications in business contexts.
This composite research portfolio highlights that the Centre is not solely concerned with AI as an abstract computational science, but actively explores how AI transforms societal sectors ranging from industry to public health.
Strategic Initiatives and National Infrastructure
The ETH AI Centre’s research impact is also shaped by strategic initiatives and collaborative programmes that extend beyond traditional academic boundaries. This section discusses selected initiatives that signify both the breadth and the influence of the Centre’s work.
The Swiss National AI Institute (SNAI), co-led by ETH Zurich and EPFL, represents a national-scale effort to coordinate AI research, infrastructure and talent development. With dedicated supercomputing resources and a mandate for open, trustworthy AI, SNAI aims to develop foundation models that align with Swiss values of transparency and openness. By incorporating expertise from over 70 AI-focused professors nationwide, SNAI will enable interdisciplinary research on foundational AI models, while advancing applications in health care, sustainability and education.
The national work on large-scale language models such as the open multilingual model Apertus underscores this commitment. Apertus harnesses a vast dataset spanning over 1,000 languages; an important step towards broader linguistic inclusion in generative AI.
The 2024 Annual Report highlights that the ETH AI Centre has deepened its network of industry partners across sectors such as cloud computing, finance, insurance and manufacturing. These partnerships serve dual purposes: providing real-world problem spaces for AI research and facilitating the translation of scholarly work into practical tools and systems. This approach exemplifies the Centre’s impact orientation, ensuring that research not only publishes new knowledge but also generates technological value for economic and social use.
Beyond formal research programmes, the Centre hosts a variety of scholarly events, including academic talk series, poster sessions and interdisciplinary workshops, that catalyse idea exchange, identify emerging trends and build community among scholars. These activities are central to fostering a vibrant research ecosystem that bridges established expertise with emerging voices.
Representative Research Case Studies
To illustrate the character and impact of the ETH AI Center’s research portfolio, we discuss several representative case studies that highlight the Centre’s scientific breadth.
The Centre for AI Value, an initiative connected to the Technology Marketing Group at ETH Zurich, was recently awarded a significant ETH research grant for a project on multimodal representation learning for retail analytics. This research aims to unify visual, textual and behavioural data using transformer-based models, addressing challenges in processing time series data and building generalisable frameworks for multimodal contexts. Such work exemplifies how AI methodologies are adapted to complex, real-world data environments and signals a growing research focus on economic applications of AI.
The development of multilingual AI models like Apertus addresses a critical gap in the global deployment of AI systems: linguistic diversity. By including underrepresented languages such as Swiss German and Romansh, this initiative demonstrates a commitment to research that acknowledges social equity and cultural complexity in AI model design.
Among the faculty at the ETH AI Center are researchers whose work focuses on human-AI collaboration, intelligence augmentation and visual analytics. These research strands go beyond autonomous decision systems to examine how humans and machines can co-construct meaning, share tasks and enhance human cognitive capabilities. This orientation aligns with broader disciplinary dialogues questioning the role of AI as a partner rather than a replacement for human expertise.
Fundamental work on robustness, reliability and safety; including exploring certification methodologies for machine learning systems, reflects a commitment to addressing well-documented weaknesses in current AI systems. By advancing techniques that ensure predictable performance under uncertainty, the Centre contributes to a crucial research frontier essential for deploying AI in high-stakes domains.
Ethical, Legal and Societal Dimensions
While the technical development of AI remains central, the ETH AI Centre explicitly incorporates ethical, legal and societal considerations into its research agenda. This commitment is manifest in multiple facets:
• The Centre promotes research on fairness, bias and explainability;
• National initiatives emphasise trustworthy and transparent models; and
• Outreach programmes engage with policy and public discourse on AI’s role in society.
This integration of ethical scrutiny with technical innovation situates the Centre within contemporary debates on responsible AI, where scholars argue that AI research must not only advance capacity but also anticipate the normative implications of deployment.
Conclusion
The ETH Zurich Artificial Intelligence Center exemplifies a comprehensive model of AI research that simultaneously advances foundational scientific understanding, fosters interdisciplinary collaboration and engages with socio-ethical dimensions of AI deployment. Its integration within national initiatives, contribution to large-scale AI infrastructure and commitment to responsible innovation position it as a leading institution in global AI research. As AI continues to evolve, the ETH AI Centre’s mission to produce trustworthy, inclusive and societally beneficial systems will remain central to shaping both scholarly discourse and practical applications.