Humans & Machines Ethics Framework: Assessing Machine Learning Influence
Humans & Machines Ethics Canvas’ main goal is to be a guide for critical thinking throughout the ethical decision-making process. It acts as a value system and an ethics framework to assess the influence of machine learning and software development while developing a system for individuals, teams, and organisations.
By Piush Vaish, Trinity College Dublin.
The recent progress in Artificial Intelligence and Machine Learning makes questions about the ethics of AI more pressing than ever. Most tech companies and their employees want to do good in the world. We all have an image of our better selves or "at our best.” We also want to avoid the risks that come with the increasing public scrutiny of ethical lapses.
Humans & Machines Ethics Canvas helps the issue of deciding on an ethical approach by addressing the following:
- What should be the base for our ethical approach?
- How should standards be applied to a specific situation?
The main idea behind Humans & Machines Ethics Canvas is to have a blueprint from ideation to successful completion of the project for individual, teams and organisations. When we make a decision, we should be able to point to the concerns, the stakeholders affected, possible actions to reduce the issue to an acceptable level and allow the proposed approach to go ahead. It is adapted from Markkula Centre for Applied Ethics, research on various frameworks from other professions such as medical, accounting, law and lean canvas.
Humans & Machines Ethics Canvas is represented as a sequence of questions and sub-questions to stimulate us beginning with the recognition and assessment of the issue, and ending with a decision and approach.
Click on the link to download a PDF file.
Concern: Recognizing an ethical issue about the product or service that our project will provide. Is there a possibility of damaging someone or a group? Does it involve a choice between good and bad alternatives?
Relevant Facts: includes the unknown and known facts. List the resources needed for the project. Is this our problem or does it belong to someone else? Is it a real problem or part of a larger one? Do we need more information? Is this a real problem or avoidance of a difficult task? Is it a human failure or technical failure?
Possible actions: List the best course of actions or abilities that will address our concerns. We want to put precautions in place or rely on any that already exist.
Stakeholders: Types of individual affected by our actions and concerns such as men/women, user/non-user, age, category etc. Organisations and groups affected such as environmental and religious groups, government agencies and competing companies.
Mental Models: is what the stakeholders believes as a result of perception, imagination and knowledge. It is psychological representations of real, hypothetical, or imaginary situations.
Externalities exist outside the recognised issue. Some examples are government regulations, law, climate impacts, privacy impacts, employment impacts and social conflicts.
Alternative Choice: List all the possible alternative course of actions. Identify any creative options.
Analysis: Evaluate to gain a better understanding. Have all the relevant persons and groups been consulted? Have we considered pros and cons for each possible choice? Which option best addresses the situation? Are we free from external influence to make this decision? Are we in a calm and unstressed state of mind? How can our decision be implemented with the greatest care and attention to the concerns of all stakeholders?
Approach: With all the information we reach a proposed approach. It can be any of the five approaches.
- The Utilitarian Approach produces the greatest balance of good over harm. An example is the ethical warfare balances the good achieved in ending terrorism with the harm done to all parties through death, injuries, and destruction.
- The Rights Approach respects the rights of all who have a stake. It includes moral rights including the rights to make one’s own choices about what kind of life to lead, to be told the truth, not to be injured, to a degree of privacy, and so on.
- The Fairness or Justice Approach treats people equally or proportionately. E.g. people get paid based on their hard work.
- The Common Good Approach serves the community as a whole, not just some members. This may be a system of laws, health care, a public educational system.
- The Virtue Approach leads to acting as the sort of person one wants to be. Honesty, courage, compassion, generosity, tolerance, love, fidelity, integrity, fairness, self-control and prudence are all examples of virtues
Reflection: Reflect on how the approach turned out and what have we learned from the specific situation? Are we willing to accept responsibility for our decision? Could we make our decision public and feel good about it? Are we comfortable with our approach? If not, retrace the steps to discover a better solution.
Values are subjective. We may not all agree to the same set of human and machine rights. We may not agree on what is the common good or even on what is a good and what is a harm. However, each approach gives us important information with which to find what is ethical in a particular circumstance.
Humans & Machines Ethics Canvas is simple and space constrained hence it gives a quick and clear vision. Viewing the history helps to understand strategic progress. Users can loop back-and-forth between steps.
A consistently rigorous approach to ethical decision-making within projects can both mitigate against expensive risks and generate new insights into products, clients, and public audience.
It would be great if you could please provide me feedback or if you like to try it out or want to contribute on the canvas, please message me on LinkedIn or leave a comment.
Bio: Piush Vaish is a student at Trinity College Dublin and a techie at heart with knowledge of applying technical standards, principles, theories and techniques in recent projects. My passion for technology drives his belief in the I.T. industry's ability and responsibility to help business to create and develop real practical opportunities.
- 3 Key Ethics Principles for Big Data and Data Science
- Ethics in Machine Learning – Summary
- Ethics In Machine Learning: What we learned from Tay chatbot fiasco?