Competence Center for Generative AI
The Competence Center for Generative AI explores the potential of algorithms (like ChatGPT or Stable Diffusion) that create new content. We research and provide information support and consulting. Our goal is to bring a better understanding of the impact of these models on society.
Our Long-Standing Expertise
With a team of experts working with generative models for language generation since 2016, the Competence Center is a hub of knowledge and expertise on these models. At the same time, we started practical work on image generation that recently converged with language inputs. Some previous works include:
- End-to-end Trainable System for Enhancing Diversity in Natural Language Generation (system description, 2017)
- Generating Restaurant Reviews from Structured Data (peer-reviewed paper, 2018)
- Headline Generation for Newspaper Articles (a project with a media partner, completed in 2019),
- Fact-aware Abstractive Text Summarization using a Pointer-Generator Network ("German Text Summarization" Challenge at SwissText 2019)
- LIHLITH - Learning to Interact with Humans by Lifelong Interaction with Humans (a fundamental European research project completed in 2020)
- Survey Paper on Evaluation of Dialogue Systems (survey paper, 2021)
- Computer vision system using medical image synthesis for data homogenization (peer-reviewed paper, 2021)
- Solution for Improving Back-Translation of Database Queries into Natural Language (peer-reviewed paper, 2022; technology integrated into the NL2DB demo)
- Automatic Generation of Meeting Summaries (Innosuisse Project, completed 2022)
- In-Depth Analysis of Automated Metrics for Text Generation Systems (peer-reviewed paper, 2022)
- Live demonstration of Swiss German speech input to stable-diffusion-based image synthesis (internal showcase, 2022)
Focus of the Competence Center
The Competence Center focuses on the following activities. We are updating these regularly, so please take a look under "Upcoming Activities".
- Webinars, talks, and presentations to inform the research and industrial communities about the capabilities, limitations, and best practices for using generative language models.
- A point of contact for the public where our experts will answer questions about ChatGPT and similar models via phone or video call (details coming soon).
- Meetups and expert meetings with the goal of knowledge exchange and cooperation encouragement among all interested parties in the field of generative language models.
- Scientifically sound experiments to assess the capabilities and limitations of ChatGPT and similar models.
- Implementation of showcases how generative language models can be used in industry, education, and research.
- A series of blog articles highlighting the latest scientific findings, industrial use cases, and guidelines for using generative language models in academia.
- Guidelines for teachers and lecturers at schools and universities on how to deal with generative language models in exercises and exams and how to benefit from them in teaching.
Join us on this exciting journey of discovery and knowledge-sharing!
We are looking forward to your participation.
- ZHAW: 02.03.2023-04.04.2023, 12:00-13:00 CET, in room TN E0.54 and Webex, a series of hybrid webinars on Generative AI: Impuls Reihe: KI in der Lehre (zhaw.ch)
- 02.03.2023: Introduction to GenerativeAI
- 09.03.2023: Demos of Generative Tools: ChatGPT, Copilot, Stable Diffusion
- 14.03.2023: Impact of Generative AI on Teaching and Exams
- 23.03.2023: Exam Design in Light of Generative AI
- 04.04.2023: Hype oder Revolution: Die Zukunft der generativen KI
- CLAIRE: 27.03.2023, 16:00-17:30 CET, Webinar: AI as a tool for Mental Health
- PHZH, ZHAW, UZH: 03.04.2023, 16:30-18:00 CET, "Opportunities and risks of AI/ChatGPT for education?" Online podium discussion: to the discussion link
- Ortsmuseum Meilen & Mindfire: 04.04.2023-23.04.2023., AI Symposium in Meilen, Ortsmuseum Meilen, Kirchgasse 14, 8706 Meilen:
- 04.04.2023, 14:30 CET: The future of AI technology
- 15.04.2023, 10:30 CET: AI & Ethics
- 23.04.2023, 14:30 CET: AI, Politics & Society
- PHZH: 19.04.2023, 18:00-19.30 CET, Continuing education "ChatGPT & Co in school": Sign Up
- PHZH and SGAB: 23.05.2023, 13:30-17:00 CET, Meeting on Automation, AI and Learning Analytics in Vocational Education and Training - Opportunities and Risk
- EATAW conference: 05.06.2023-07.06.2023, Winterthur: various applications of generative AI text production tools will be discussed in workshops, research papers and teaching orientated presentations. Focus is the support of academic writing and its instruction.
- 24.03.2023 / SwissLife / “All-rounder ChatGPT: How self-determined is artificial intelligence?”, Interviewee: Prof. Thilo Stadelmann, ZHAW
- 01.03.2023 / 20 Minuten / “Wegen KI-Schummeleien wollen Unis Studierende mehr mündlich prüfen”, Interviewee: Prof. Thilo Stadelmann, ZHAW
- 20.02.2023 / 20 Minuten / “These jobs will be eliminated by AI in the future”, Interviewee: Prof. Thilo Stadelmann, ZHAW
- 17.02.2023 / Schaffhauser Nachrichten / “Will we journalists even be needed in the future?”, Interviewee: Prof. Thilo Stadelmann, ZHAW
- 13.02.2023 / Finanz und Wirtschaft / “ChatGPT heralds tech revolution”, Pascal Kaufmann, Lab42; Prof. Thilo Stadelmann, ZHAW; Prof. Benjamin Grewe, ETH
- 11.02.2023 / NZZ Magazin / “Artificial intelligence: Microsoft can only win, Google can only lose”, quoted: Prof. Thilo Stadelmann, ZHAW
- 04.02.2023 / SRF Wissenschaftsmagazin / “The search for a good handling of the language software ChatGPT at universities.”, Interviewee: Prof. Mark Cieliebak, ZHAW
- 06.09.2022 / NZZ Folio / “The computer is catching up with us: Machines carry on conversations and create images - with dramatic consequences”, responsible for GPT-3 implementation: Jan Deriu, ZHAW