GEOG 495 A: Special Topics

Instructor: Bo Zhao ｜ Course Duration: 4 Weeks

This Course explores the intersection of artificial intelligence (AI) and geographic knowledge through the lens of machine learning, deep learning, and generative models. Designed for students in the social sciences and humanities, this course introduces foundational AI concepts and their applications in geospatial analysis, such as land use mapping, intelligent agents, and spatial storytelling.

Through a mix of asynchronous lectures, hands-on labs, self-paced learning, and collaborative discussion, students will experiment with tools like ChatGPT, the Segment Anything Model (SAM), and Python-based mapping libraries. By the end of the course, students will be equipped to critically evaluate and creatively apply AI tools to real-world spatial problems — from urban inequality to cultural heritage preservation — and build their own geospatial AI prototypes.

No advanced programming background is required. Curiosity about AI, maps, and their implications in society is encouraged.

Format:

• 🕒 Asynchronous recorded lectures

• 🧪 Weekly hands-on Labs (via Google Colab + ArcGIS Pro)

• 🧠 Self-paced learning modules (YouTube / Hugging Face Spaces)

• 💬 Discussion Board participation

🎯Objectives

• Introduce AI, Machine Learning, and Generative AI with a geospatial focus

• Explore visual (e.g., SAM) and language-based (e.g., ChatGPT) AI tools

• Build spatially aware applications and intelligent agents

• Encourage ethical reflection and critical thinking about AI's role in geography

🗓️ 4-Week Course Structure

✅ Week 1: Foundations of Geospatial Machine Learning

Topics:

• AI, ML, and DL fundamentals in geography

• Vector vs. raster data; spatial modeling

• ML techniques: regression, decision trees, clustering

• Use cases: spatial prediction in housing, demographics

Lab 1: Google Colab + ArcGIS Pro: Build an ML model (e.g., decision tree) to analyze geospatial data such as housing prices or urban growth

Activities:

• Self-learning: YouTube/LinkedIn “ML for Maps”

• Code practice with Scikit-learn and Geopandas

• Discussion: “What kind of knowledge can spatial ML reveal?”

✅ Week 2: Deep Learning & SAM in Remote Sensing and Spatial Analysis

Topics:

• Convolutional Neural Networks (CNNs) for spatial imagery

• Segment Anything Model (SAM) and its applications

• segment-geospatial Python library + ArcGIS workflows

• Comparing visual and language models for spatial data

Lab 2: Hugging Face + Colab: Segment trees, pools, buildings from satellite images using SAM

Activities:

• Try ESRI’s SAM demo

• Run segmentation tasks using real-world imagery

• Discussion: “Where do image-based models succeed or fail in geospatial contexts?”

✅ Week 3: Large Language Models and Geo-Intelligent Agents

Topics:

• LLMs: ChatGPT, LLaMA, and spatial knowledge

• Designing chatbots for geospatial tasks

• Prompt engineering, API calling, Whisper for transcription

• Ethical considerations in AI dialogue and spatial bias

Lab 3: Colab + API: Build a location-aware chatbot that provides spatial guidance (e.g., historical site tours, natural disaster Q&A)

Activities:

• Use OpenAI or LLaMA via API

• Practice Whisper for speech-to-text of spatial interviews

• Discussion: “What would a ‘geo-intelligent agent’ look like in practice?”

✅ Week 4: Spatial AI for Social Impact + Final Project

Topics:

• AI for social good: environment, justice, disaster response

• Generative AI for historical storytelling, spatial narratives

• Final project guidance + design thinking

• Reflecting on future roles of AI in geography

Lab 4: Final Project Development: Design and present a spatial AI application (e.g., ChatGPT assistant, SAM land use mapper, food access predictor)

Activities:

• Peer feedback on project ideas

• GitHub repo & Colab notebook setup

• Discussion: “How will you shape the future of spatial AI?”

🧰 Required Tools

• Google Colab

• GitHub account

• Hugging Face account

• OpenAI API key (free tier)

• Access to ArcGIS Pro (via UW license)