The AI for Good Specialization on Coursera, led by AI expert Robert Monarch, is a comprehensive program that explores the application of artificial intelligence in addressing some of the world's most pressing challenges. The specialization consists of three courses:

1. AI and Public Health: This course introduces learners to the concept of using AI to address social and environmental issues. It provides a comprehensive understanding of how AI can be used to improve public health outcomes.

2. AI and Climate Change: This course provides a detailed overview of anthropogenic climate change, its potential impacts, and how AI can be a powerful tool in mitigating these effects. It explores the intersection of AI and climate science, offering insights into how technology can be harnessed for environmental sustainability.

3. AI and Disaster Management: This course delves into the role of AI in both the prediction and management of natural disasters. It offers a deep dive into the use of AI in predicting extreme weather events, managing disaster response, and planning for long-term recovery.

Each course in the specialization is designed to be accessible to learners of all backgrounds and includes a variety of real-world case studies. The specialization offers a unique perspective on the role of AI in creating a positive societal impact and is a valuable resource for anyone interested in the practical applications of AI.

My Certificate

"AI for Good" is a term used for AI that provides support and prevents, mitigates, or resolves problems affecting human life or the environment. In this specialization, we focus on how it does this in the environment, health, justice, and humanitarian action sectors.

AI applications make decisions based on historical data. If the data is biased, the AI will be biased.

As an example of how to think about AI for Good, here are the UN Sustainable Development Goals:

1. No Poverty
2. Zero Hunger
3. Good Health and Well-being
4. Quality Education
5. Gender Equality
6. Clean Water and Sanitation
7. Affordable and Clean Energy
8. Decent Work and Economic Growth
9. Industry, Innovation and Infrastructure
10. Reduced Inequality
11. Sustainable Cities and Communities
12. Responsible Consumption and Production
13. Climate Action
14. Life Below Water
15. Life on Land
16. Peache, Justice and Strong Institutions
17. Partnerships for the Goals

"Do no harm" is a good principle meaning everyone impacted by a project is left improved or unchanged.

Deep Learning is a subset of Machine Learning, which is a subset of AI.

Areas of concern to consider when applying AI to a problem:

• Is AI the right solution? This needs to be figured out early so that time and resources are not wasted.
• Personally identifiable information (PPI) should be treated with care and most likely not stored. This data should not be published or archived since it could be used to oppress certain groups of people.
  • An example is not publishing the location of endangered species, since poachers could use this information to find and kill them.
• Consider the advice from all stakeholders that would be impacted by the project.

A good problem statement should:

1. Be clear, concise, and specific
2. Identify key stakeholders
3. Give an idea of what success looks like
4. Not mention the specific technology you aim to deploy

At the end of the "Explore" phase, we should be able to answer the following questions:

• What is the specific problem we are addressing?
• Who are the stakeholders?
• Do we have access to or can we collect the data we need?
• Could AI add value? Where and how specifically?
• How does the "do no harm" principle come into play?

After the "Design" phase, we should be able to answer the following questions:

• How will we address issues with imbalances, biases, privacy, or other concerns with your data?
• What kind of model will you implement, and how will you measure its performance?
• How will your design address the problem you set out to work on?
• how will the end user interact with your system?

After the "Implement" phase, we should be able to answer the following questions:

• Is our model performance acceptable?
• Are end users able to successfully use our system?

• The preservation and restoration of biodiversity is a key component of climate change mitigation and adaptation.
• Some experts suggest that we are currently in the midst of the sixth mass extinction event in the history of the planet and that humans play a major role in causing it.

We do not have to train the model from scratch in each case. In this case, we could use a pre-trained model already trained by Microsoft (MegaDetector) and simply fine-tune it to our specific use case.

A disaster is a sudden, calamitous event that seriously disrupts the functioning of a community or society and causes human, material, and economic or environmental losses that exceed the community’s or society’s ability to cope using its resources.

Laboratory solutions should not be deployed during a disaster or shortly after. This is because the laboratory environment is very different from the real world and the model will not perform as well as it did in the lab.

Work that does not harm is most likely to succeed if done in the recovery, preparation, or mitigation phases, instead of the response phase.

1. Work on general-purpose technologies to help communities help themselves.
2. Support low-resource languages with better technologies like translation and search.
3. Default to private data practices. Aggregate data and ML models can amplify privacy risks.
4. Avoid projects that involve analysis of social media data and work funded by oppressive governments.
5. Engage with impacted communities.

Overhead imagery is a good source of data for disaster management. It can be used to identify damaged buildings, flooded areas, and other useful information.

• Satellites
• Planes
• Drones

All material derived from the AI for Good Specialization on Coursera.