Principled AI at Amii
Our approach to the ethical development of AI technologies
Our approach to the ethical development of AI technologies
Principled AI at Amii includes all of the responsibilities, practices and initiatives we undertake to ensure our applied AI work is completed under an ethical framework.
We care deeply about how AI and ML research is ethically developed, adopted by industry, and taught to students, professionals and broader society.
In practice, we align our work to benefit society.
All United Nations Member states adopted the 2030 Agenda for Sustainable Development in 2015, which provides a shared blueprint for peace and prosperity for people and the planet, now and into the future.
Our work aligns with the spirit and intent of the 17 sustainable development goals.
When engaged with industry, we proactively evaluate opportunities to prevent negative impacts on our collective progress.
The sustainable development goals are No Poverty; Zero Hunger; Good Health and Well-Being; Quality Education; Gender Equality; Clean Water and Sanitization; Affordable and Clean Energy; Decent Work and Economic Growth; Industry, Innovation and Infrastructure; Reduced Inequalities; Sustainable Cities and Communities; Responsible Consumption and Production; Climate Action; Life Below Water; Life on Land; Peace, Justice, and Strong Institutions; Partnerships for the Goals.
When we work with industry, the purpose of the project does not need to be to advance one or more of these goals forward. We use the goals as a lens to better understand where the project could have intended and unintended benefits or consequences.
In all of our work, we demonstrate a commitment to an ethics and rights-based approach to developing AI knowledge, talent and applications.
It is insufficient to signal our intent to leverage AI for societal benefit.
We build, use and audit our AI systems, processes and documentation to ensure that we act in accordance with our principles and values in three categories:
With concerns about AI bias already impacting individuals globally, Fairness and Non-discrimination principles call for AI systems to be designed and used to maximize fairness and promote inclusivity.
Amii bases its understanding of fairness and non-discrimination on the broad view approach in the paper A Framework for Understanding the Unintended Consequences of Machine Learning. In this paper, Suresh and Guttag “argue that analyzing the consequences of a particular algorithm should begin with a thorough understanding of the data generation and ML pipeline that led to its output.” These include six biases: historical, representation, measurement, aggregation, evaluation, and deployment.”
AI systems should respect individuals’ privacy while using data to develop technological systems and provide impacted people with agency over their data and its decisions.
AI systems should be safe, perform as intended, secure, and resistant to being compromised by unauthorized parties.
Individuals play a vital role in the development and deployment of AI systems and the systems’ impacts. Our teams are called on to use their professionalism and integrity to ensure that the appropriate stakeholders are consulted and that long-term effects are considered and mitigated.
At Amii, our team accountability and professional responsibility is to execute all of the Principled AI tasks assigned to you promptly and appropriately.
Once we move into hands-on ML work with an industry partner, the requirements of our Principled AI Framework become much more tactical and customized from project to project.
For all of our projects that included hands-on ML work, there are three tools that serve as communication tools that showcase how we are collaboratively mitigating Principled AI risk.
These tools serve as proof points that we acted according to our responsibilities throughout the project.
We act as a convener and collaborator, upholding high standards of ethical excellence across the entire AI ecosystem.
In our unique role bridging academia and industry, we advance the world’s knowledge in AI with diligence and rigour. We continually work with a wide range of partners to challenge our understanding of best practices, promote thought leadership, and draw from the wealth of knowledge of our peers from multidisciplinary backgrounds who have explored and grappled with complex ethical landscapes.
You can find conversations around Principled AI at Upper Bound, TechAid, AI Meetup, AI Seminar, and our many external speaking engagements year-round.
This policy is effective as of April 1, 2023.
If you have any questions about our approach to Principled AI, please contact us.
Looking to build AI capacity? Need a speaker at your event?