How do you measure the social and environmental impact of your AI projects?

2 Assess your data

Data is the fuel of AI, but it can also be the source of bias, error, or harm. You need to assess the quality, quantity, and diversity of your data and ensure that it is relevant, accurate, and representative of your target population. You also need to respect the privacy, consent, and rights of the data providers and users. You can use tools such as Data Ethics Canvas or Data Nutrition Project to help you with data assessment.

3 Monitor your model

Once you have built your AI model, you need to monitor its performance and behavior. You need to test your model for accuracy, reliability, fairness, and explainability. You need to identify and mitigate any potential errors, biases, or harms that your model might cause or amplify. You need to document your methods, assumptions, and limitations and communicate them transparently to your users and stakeholders. You can use tools such as AI Fairness 360 or LIME to help you with model monitoring.

4 Evaluate your impact

After you have deployed your AI model, you need to evaluate its impact on the real world. You need to measure the outcomes and effects of your model on your users, stakeholders, and society at large. You need to consider the positive and negative impacts, the intended and unintended impacts, and the short-term and long-term impacts. You need to collect feedback and evidence and use them to improve your model and address any issues or concerns. You can use tools such as Impact Management Project or AI Impact Assessment to help you with impact evaluation.

5 Engage with others

You cannot measure the social and environmental impact of your AI projects alone. You need to engage with others who have different perspectives, experiences, and expertise. You need to collaborate with your team members, partners, and clients and ensure that they share your goals and values. You need to consult with your users, stakeholders, and affected communities and involve them in your design and evaluation processes. You need to learn from other computer scientists, researchers, and practitioners who are working on similar or related problems and solutions.

6 Follow the guidelines

Finally, you need to follow the guidelines and standards that have been developed by various organizations and institutions to promote ethical and responsible AI. These guidelines provide you with best practices, principles, and frameworks to help you measure the social and environmental impact of your AI projects. Some examples of these guidelines are the OECD Principles on AI, the IEEE Ethically Aligned Design, and the UNI Global Union Top 10 Principles for Ethical AI.

7 Here’s what else to consider

This is a space to share examples, stories, or insights that don’t fit into any of the previous sections. What else would you like to add?