Welcome to the AI SandboxThe AI Sandbox is a weekly, hands-on learning space where faculty, researchers, and staff across campus can explore and experiment with real AI tools. Each session introduces models used in research, from image segmentation to text classification, and walks you through how to run them in your browser, step by step. No prior experience is required, just bring your laptop and curiosity!These are not lectures or office hours. You’ll follow guided demos, test out pre-trained models, and have time to explore either with example datasets or your own research material. Each week focuses on a different theme, with support provided throughout.Due to popular demand, registration is now required. Sessions are small to ensure everyone gets guided support, but you’re welcome to bring along colleagues; groups of up to five can attend together.

Next Session: Introduction to Natural Language Processing AI models do more than read text. They can summarize, translate, compare, and classify it.In the upcoming AI Sandbox sessions, you’ll see how modern language models work and how they can support common research tasks, from reviewing abstracts to analyzing interview transcripts.You’ll get hands-on practice with:EmbeddingGemma-300M for measuring how similar two pieces of text are, useful for clustering, deduplication, and alignment checksBART-Large-MNLI for classifying text into your own categories without extra trainingMADLAD-400 for straightforward translation across languagesBART-Large-CNN for turning long passages into concise summariesThe Green Court building at 3520, Green Ct., MI 48105. The MIDAS offices are on the 3rd floor.The west side of the building is reserved for non-University of Michigan use. There is a U-M yellow lot immediately to the east of the building. Please do not park in the Black & Veatch parking lot.