Artificial Intelligence (AI) is rapidly expanding into the health care environment. AI tools like automated decision-making systems, generative AI systems, and ambient listening-based charting tools have the potential to enhance clinical education, reduce clinical errors through predictive models, and automate routine tasks to save time and streamline workflow. On the flip side, AI use presents risks, such as “hallucinations,” bias, information provided without evidence, user over-reliance, and skill erosion, among others. The principles of the Professional Standards for Midwives outline the expectations of midwives in all areas of their work; these principles extend to the use of AI.
Relevant Professional Standards
- 2. Be competent in all areas of your practice.
- 32. Conduct yourself in a way that promotes clients’ trust in you and the public’s trust in the midwifery profession.
- 37. Take every reasonable precaution to protect the confidentiality and privacy of your clients’ personal health information, unless release of information is required or permitted by law.
Professional Knowledge and Practice
When it comes to professional knowledge and practice, midwives should consider the risk that is posed by each AI tool they utilize. Risks increase, for example, when AI tools have no “human in the loop” features, which means that the midwife is not involved at any point in the AI workflow to ensure accuracy, accountability, safety, or ethical processes. Risk also increases if the AI tool directly influences clinical decision-making or formulates a diagnosis or utilizes a clients personal health information. It is the midwife’s responsibility to get to know the tools well enough to be satisfied that the AI will not interfere with their ability to meet their professional standards. AI tools should never be used to override clinical judgement, decision-making, or care-planning; midwives must use their own skills to provide midwifery care and be competent in all areas of care. AI should never be used as a substitute for care nor to provide care that a midwife is not otherwise competent to provide.
Integrity
Transparency is the consistent practice of openness and honesty in communication and decision-making. It means sharing information clearly and completely – without hiding essential facts – to build trust and credibility in relationships. Being honest with your clients about which AI tools you use, when they are used, how information and data may be collected, and how it is stored will allow clients to make informed decisions about their consent for its use in their care. Providing clients the ability to consent to (or decline) the use of AI tools is an essential step in providing ethical care. It’s also a midwife’s responsibility to provide reasonable accommodation, when possible, to their clients if they express a desire for no or minimal involvement of AI in their care.
There are also risks related to private health information becoming part of Large Language Models, or LLMs, where personal details are used as training data to understand, summarize, translate, or generate human-like language. It is imperative that midwives continue to protect personal health information and privacy of their clients and continue to abide by the Personal Health Information Privacy Act, 2004 (PHIPA). Common tools like Gemini, ChatGPT, and Claude rely on constant feeds of information to provide materials through their LLM programs; it is the midwife’s responsibility to never input a client’s personal information into any of these types of programs.
Midwives: Call to Action
Using AI well in your day-to-day work may help reduce administrative tasks and free up more quality time with clients. At the same time, AI comes with real risks. If those risks are not understood and managed, they may make it harder to meet your professional standards.
The College advises midwives to:
- Get training before using AI tools. Make sure you understand what the tool can and cannot do, including its limits and risks.
- Review all AI-generated material carefully. Check that it is accurate, complete, and free from bias before relying on it.
- Use extra care before adding AI-generated material to a client record or using it to support clinical decisions. Assess it first, using your own professional judgement.
- Keep a clear record of how AI was used so the material can be reviewed or audited later.
- Take all reasonable steps to protect the confidentiality and privacy of clients’ personal health information.
- Provide care in an inclusive way. Clients who choose not to use AI-based tools may need reasonable accommodations.