This ethical model checklist was developed by Amy Shah, MD, MBA, MS to help individuals and small groups determine if AI used for their project or work is beneficial and uses ethical principles. The first item in the checklist is to determine the goal of the work/project and if AI is suitable for that goal.
Goal: What is the goal that I will be using AI for in my project/work?
To the best of my ability, can I determine if the AI was developed with ethical principles in mind?
Explainability - Is the AI transparent and clear about what inputs and decision making that
went into its recommendations?
Fairness - Does this AI appear to treat individuals and/or groups of individuals equitably?
Robustness - Does this AI minimize security risks and prevent security attacks that might
cause private information to be leaked?
Transparency - Can individuals and groups see how the AI works and evaluate its function?
Can this AI be trusted or be trustworthy?
Privacy- Does this AI prioritize and safeguard consumers’ privacy and data rights and provide
explicit assurances to users about how their personal data will be used and protected?
To the best of my ability, can I determine if the AI treats individuals and groups in an ethical manner?
Integrity - Does the AI have consistency of values and actions so that what is advised and what is done are in harmony?
Respect - Does the AI acknowledge the inherent value of all individuals and treat all with dignity?
Responsibility - Can the use of this AI be considered responsible? Are we able to have awareness that our actions have consequences that can be positive or negative? Is there a way we could be held accountable for the actions of this AI?
Fairness - Are the ethical decisions of the AI unbiased? Do the actions of the AI treat everyone
equally regardless of status, position, or personal attributes?
Compassion - Does the use of the AI account for empathy and understanding towards others? Do all stakeholders have equal say? Are we keeping in mind that there is a human element in every decision we make?
Courage - Does use of AI allow for respectful disagreement? Courage means doing what is right despite the idea being risky or unpopular. It means standing up for what you believe in. Is there room for dissenting thought in your project or goal?
Wisdom - Have you made AI related decisions based on being informed on the issues and facts while also taking into account the stakeholders?
To the best of my ability, have I determined if AI involvement will affect human medical decision making? If so, have I considered the following principles?
Non-maleficence - Does the AI cause any harm to individuals or groups? Non-
maleficence means “do no harm.”
Beneficence -Does the AI actively help others welfare and health outcomes?
Health maximization - Does the AI allow an environment that maximizes health for the
individual and opportunities for health of the public?
Efficiency - Does the AI use scarce medical resources efficiently?
Respect for autonomy - Does the AI allow for the patient’s right to determine what will happen to the and encourage their ability to make their own medical choices.
Justice - Does the AI allow for equity for all people so that all people can have health care at the same quality?
Proportionality - Does the use of AI weigh the individual needs versus the needs of the greater good in a beneficial and equitable way?
This checklist for individuals and groups is meant to help with taking careful consideration about the safety of AI in one’s work project or goal. This was developed based on the following reference materials. For more information, please see this article: https://hackernoon.com/human-mental-health-and-artificial-intelligence
References:
- Peter Schröder-Bäck, et al. Teaching seven principles for public health ethics: towards a curriculum for a short course on ethics in public health programmes. Schröder-Bäck et al. BMC Medical Ethics 2014, 15:73 http://www.biomedcentral.com/1472-6939/15/73
- https://axies.digital/ethical-decision-making/
- https://www.ibm.com/policy/ibm-artificial-intelligence-pillars/
- Images created by DALL-E/ChatGPT