Generative AI (Gen AI) is enhancing patient experiences and health outcomes while reducing administrative tasks for healthcare professionals. As the demand for healthcare services grows, organizations are increasingly turning to AI tools and automation for support.
However, implementing AI technologies is often not straightforward. Below are some of the most common challenges organizations face, along with suggestions for addressing them:
1. Lack of Trust
Is your team familiar with the capabilities of Gen AI? A significant concern is that even if AI is successfully implemented, both employees and patients may hesitate to trust it. This skepticism is understandable, as Gen AI and agentic AI are relatively new technologies, and many may lack the necessary information to feel confident in their use.
Trust is context-dependent. For instance, while patients and providers may be reluctant to allow AI to make critical patient-care decisions, they might be more comfortable with AI summarizing clinical notes or providing decision support. It’s essential to align AI use cases with governance policies and organizational goals. Throughout the process, users must remain accountable for any AI-generated content or recommendations.
A notable initiative aimed at fostering trust in AI within healthcare is the Coalition for Health AI (CHAI™), which works to establish standards for responsible AI use and offers guidance to healthcare systems at various stages of their AI journey.
2. Concerns About Accuracy
When utilizing Gen AI for information dissemination, it is crucial to ensure that the input data is accurate and that reliable algorithms are in place to achieve optimal outcomes. Whether AI is used for information provision, content generation, recommendations, or actions, human oversight and ongoing monitoring are necessary to maintain accuracy and trust among stakeholders.
The level of risk associated with a specific use case will dictate the required monitoring and oversight. Many Gen AI tools have improved their ability to reference sources, facilitating the verification of generated content. Human evaluation remains essential.
3. Staff Training
The introduction of AI tools can provoke a range of reactions based on perceived risks and unfamiliarity. Addressing these fears through education and guidance is vital. Just as one would read a manual before operating a new vehicle, teams expected to use AI should receive proper training. Policies governing AI use should be developed with input from stakeholders across the organization.
AI applications involving patients must be carefully monitored and evaluated for risks to minimize potential negative outcomes. Patient care should always be prioritized. Some use cases may need to be avoided due to discomfort among key stakeholders, while others may offer sufficient benefits to warrant close supervision.
Healthcare staff should have realistic expectations regarding AI capabilities and be equipped to communicate these expectations clearly to all stakeholders.
4. Protection of Intellectual Property
Similar to how drivers must adhere to traffic laws, AI users should implement policies to guide the use of copyrighted materials. Concerns typically revolve around two main issues: (1) the risk of unintentionally infringing on existing copyrights and (2) the ability to copyright new material generated by AI. Users should consult their organization’s legal team when creating any content that may be considered intellectual property. Generally, AI-generated content should be viewed as preliminary drafts that require user review and modification to avoid duplicating existing works.
Each of these challenges requires careful consideration and proactive measures to mitigate risks. Establishing a comprehensive AI governance program, complete with training and policies, can facilitate responsible and effective AI implementation.
For further insights, visit Human-centric AI research: A study in employee trust and workplace experience.