Learn AI's role in addressing complex challenges. Build skills combining human and machine intelligence for positive real-world impact using AI
The AI for Good Specialization showcases how AI can be part of the solution when it comes to addressing some of the world’s biggest challenges in areas like public health, climate change, and disaster management.
In these courses, you’ll learn from instructor Robert Monarch, who has over 20 years of experience building AI products in industry and working at the intersection of AI and public health and disaster management. Robert is also the author of Human-in-the-Loop Machine Learning, a book focused on human-centered AI applications.
Throughout the courses, you'll hear from experts working on AI for Good initiatives aimed at addressing social and environmental issues. By combining human and machine intelligence, real-world datasets, best practices around data privacy, and ethical considerations, you’ll develop the knowledge and fundamental skills to tackle your own AI for good projects.
- Create an air quality monitoring application for the city of Bogotá, Colombia
- Explain the mechanisms behind anthropogenic climate change and the ways it is impacting global temperatures and weather patterns.
- Identify ways in which AI techniques used in the context of addressing climate change, and use the techniques.
- Frame the role of AI in fighting climate change, the types of AI approaches that can be used, and the factors to consider when deploying AI in climate contexts.
- Explore a real-world wind power forecasting problem using the AI for Good framework
- Use satellite imagery from Hurricane Harvey in 2017 to work through the entire AI for Good project pipeline.
The AI and Public Health course is designed to introduce learners to the concept of using artificial intelligence to address social and environmental issues.
We'll start by defining AI for Good, and explore various examples of AI for Good projects. You will get hands-on with a real-world concern, air quality in Bogotá, and use the AI for Good framework to define the problem, identify stakeholders, and determine where AI could fit. You will also learn how to interpret exploratory data analysis (EDA) graphs and use Python to explore air-quality data. The course also covers approaches for missing data imputation, model performance evaluation using mean absolute error (MAE), and interpreting EDA graphs and heatmaps.
This is the first course in the AI for Good Specialization, which demonstrates how AI is being utilized to address some of the world’s biggest challenges — and provides a framework for you to be part of the solution.
This is a beginner-friendly course. Learners should know high school-level math and basic spreadsheet operations.
- Summarize the importance of data for an AI project
- List various examples of machine learning algorithms
- Define AI for Good and identify AI for Good examples
- Explain at a high level what artificial intelligence, machine learning, and deep learning are, and their relationship with one another.
- Describe some AI limitations, concerns and ethical questions surrounding it
- Identify the key components of supervised learning
- Describe the AI for Good project development framework
- List the expected outcomes from every phase of the framework
- Recognize how the AI for good framework is applied in real-world projects
- Explore a real-world problem (air quality in Bogotá) using the AI for Good framework
- Interpret common exploratory data analysis (EDA) graphs
- Use a Jupyter Notebook to run Python code to explore air-quality data
Lab: Explore Phase - Exploring Air Quality Data
- Clarify how to approach to an AI problem and why
- List various challenges you may encounter in such AI related project
- Summarize the tasks you do in the design and implement phase
- Describe some approaches for missing data imputation
- Determine the models performance using MAE
- Differentiate between models based on their performance using MAE
- Interpret common exploratory data analysis (EDA) graphs and heatmaps
- Design and Implement the AI4G project, including the model strategy and user experience
- Determine how to ensure data protection and privacy
Lab 1: Design Phase - Estimating Missing PM2.5 Values
Lab 2: Design and Implement Phase - Estimating Between Sensors and Constructing a Map
Quiz: Air Quality Design and Implementation
In Course 2: AI and Climate Change, you will begin by learning the basics of anthropogenic climate change — why it is happening, its projected impacts, and how it is already driving extreme weather around the globe. You will then learn how machine learning techniques can lessen climate change’s impacts and help communities prepare for those that do occur.
Learners will take part in two hands-on labs. In the first, you will use data modeling techniques to visualize how climate change is modeled to cause average temperatures to change in different locations around the world. In the second, you will build a model that forecasts how much power wind turbines in different locations will generate.
This course is part of the AI for Good Specialization, which demonstrates how AI is being harnessed to tackle some of the world’s biggest challenges — and provides a framework for you to be part of the solution.
This is a beginner-friendly course. Learners should be familiar with high school-level mathematics and basic spreadsheet operations. It is recommended that learners first complete Course 1: AI and Public Health.
- Describe how increased GHG emissions are creating rising temperatures on Earth (i.e. greenhouse effect).
- Describe ways in which climate change is causing social or environmental crises.
- Assess the applicability/relevance of AI technologies for wind energy forecasting.
- Identify contexts and problems in which AI has been used and can be used in the context of climate change.
Lab: Exploring Global Temperature Change
Quiz: Climate Change & Global Warming
- Assess the applicability/relevance of AI technologies for wind energy forecasting.
- Determine how machine learning approaches (e.g. LSTMs) can be used for predicting wind energy generation.
- Determine what metrics are helpful to measure the performance of a regression model.
Lab 1: Explore Phase - Distribution of the Wind Power Data
Lab 2: Design Phase - Feature Engineering on the Wind Power Data
Lab 3: Design Phase - Forecasting Wind Power 24 Hours in Advance
- Describe the impact of climate change on habitat loss and diversity loss.
- Determine how biodiversity loss and climate change are related.
- Explain how and why image data is important in the fight against climate change.
- Use different techniques for processing and transforming image data.
- Determine how camera trap data is currently analyzed by scientists and what kind of automatic analysis would be helpful to them.
Lab: Explore Phase - Exploring the Karoo Image Data
- Define what Convolutional Neural Networks (CNNs) are.
- Implement a CNN for classifying different types of animals on camera trap data.
- Define the difference between training from scratch and fine-tuning a pre-trained model.
- Demonstrate the utility of using models trained on general-purpose data that can be fine-tuned on domain-specific datasets.
- Evaluate how the model does using confusion matrices and derivative metric(s).
Lab 1: Design Phase - Using the MegaDetector
Lab 2: Design Phase - Fine-Tuning Your Classification Model
Lab 3: Implement Phase - Object Detection Pipeline
In Course 3, AI and Disaster Management, you will begin by learning how natural disasters create both short and long-term impacts on the economy, environment, and community. You will learn how the disaster management cycle can be used to reduce the impacts of a disaster and guidelines for how you can engage in disaster-related projects.
You will get hands-on experience through a pair of case studies. The first case study looks at Hurricane Harvey, which devastated parts of the Caribbean, Mexico, and the southern United States in 2017. You will use computer vision to identify hurricane damage from satellite imagery. The second focuses on the catastrophic 2010 earthquake in Haiti. You will use natural language processing (NLP) to compare requests for aid that victims made shortly after the earthquake with those that they made several months later.
This course is part of the AI for Good Specialization, which demonstrates how AI is being harnessed to tackle some of the world’s biggest challenges — and provides a framework that you can apply to any AI for Good project in assessing if AI can add value.
This is a beginner-friendly course. Learners should know high school-level math and basic spreadsheet operations. It is recommended that learners first complete Course 1: AI and Public Health and Course 2: AI and Climate Change.
- Examine the immediate and long-term impacts disasters have on communities.
- Define the four phases of the disaster management cycle and the actions involved at each phase.
- Describe ethical considerations and leadership guidelines when working with communities affected by disasters.
Quiz: AI and Disaster Management
- Explore satellite images from Hurricane Harvey in 2017 to define the problem, identify the stakeholders of your project, determine where AI could fit and whether it is necessary or not, explore the data of the project, and design the AI4G project.
- Define a use case when satellite data can provide precious resources to guide disaster response
- Identify ethical and privacy constraints when working with aerial imagery in the aftermath of a disaster.
- Implement a CNN for classifying satellite images
- Evaluate how the model does using confusion matrices and derivative metric(s)
- Compare the advantages and disadvantages of using imagery from satellite, aircraft, or drones.
Lab 1: Explore Phase - Exploring the Hurricane Harvey Satellite Image Data
Lab 2: Design Phase - Classifying Images with a Convolutional Neural Network
Lab 3: Implement Phase - Analyzing Classification Features and Develop a Geo-Locator
Quiz: Damage Assessment for Disaster Response and Recovery
- Explore the Haiti Earthquake disaster from 2010 to define the problem, identify the stakeholders of your project, determine where AI could fit and whether it is necessary or not, explore the data of the project, and design the AI4G project.
- Describe how to process text data for natural language applications.
- Implement LDA for topic modeling and assess its performance using coherence metric.
Lab 1: Explore Phase - Exploring the Haiti Earthquake Text Message Data
Lab 2: Design Phase - Cleaning and Processing Text Data
Lab 3: Design Phase - Performing Topic Modeling on Text Messages with LDA