Overcoming generative AI implementation blind spots in health care
As health care executives integrate generative AI into their workflows, our survey shows that a sole focus on data could leave them blindsided by other vital success factors.
Asif Dhar
Jay Bhatt
Michael McCallen
Tom Hittinger
Maulesh Shukla
Hemnabh Varia
Generative AI has elicited buzz across various industries, and health care is no outlier. Many health care organizations see generative AI’s promise in achieving greater efficiency, effectiveness, and innovation from code to cure, and are planning to accelerate AI investments this year. Additionally, consumers have shown that they’re willing to engage with generative AI on their health care journeys.1 As health care organizations begin to integrate generative AI into their workflows, taking a holistic, institutional approach may help achieve a successful implementation at an enterprise level.
Findings from the Deloitte 2024 Health Care Generative AI Outlook survey of 60 health care executives suggest that a traditional data-focused approach to implementing generative AI (gen AI) could be too narrow, highlighting the potential need for a broader strategy. We found that over 70% of executives are highly focused on data considerations like data availability, quality, compliance, security, and privacy during implementation (see figure). However, potential blind spots (which we’ve defined as considerations that less than 60% of executives are focused on) may prevent health care organizations from successfully integrating gen AI into their workflows. Let’s take a closer look at these low-focus areas that may be blind spots for leaders:
- Effective governance is lost among other data priorities. Implementing a governance model, inclusive of data, is important to help ensure the effective use and quality of data, mitigate data bias for equitable design, and safeguard patient privacy. Each of these factors can help build both consumer and employee trust. Despite the potential benefits, creating a data governance model (60%) and mitigating data biases (45%) rank among the lowest considerations for health care executives.
- Health care leaders aren’t paying enough attention to what matters most to consumers. Compared to the traditionally high focus on data considerations, the surveyed executives said they’re less focused on building consumer trust in gen AI and improving data-sharing (50%) and educating patients about AI and its risks (45%). Our earlier research showed many consumers are already engaging with gen AI for their health care decisions, but they seek greater transparency on how their data is used and who’s using it. With less of a focus on what’s important to the consumers, health care organizations may find that trust and engagement levels drop.
- Investing in and responding to workforce needs remain low priorities. Gen AI technology has understandably sparked concerns about the potential effect on the workforce. Our recent cross-industry research shows early AI adopters see more value in using the technology to upskill and reskill their employees than reducing costs by eliminating jobs. However, the survey findings show that health care executives may miss this imperative, as workforce upskilling (63%), addressing their concerns and reassuring trust (60%), and change management in terms of shifting job roles and workforce composition (57%) ranked among the lowest considerations for successful implementation of gen AI.
While gen AI has the potential to make a transformative impact akin to that of personal computers and the internet, technological innovations like these often take decades to become ubiquitous. Personal computers, introduced in the 1970s, only became household staples by the late 1990s, and the internet, born in the 1960s, reached mainstream usage in the 1990s.2 Recent emerging technologies such as AI, cloud computing, and others have seen limited adoption outside of large organizations typically driven by automation.3 To avoid a long and slow ramp-up, health care organizations should focus on multiple factors to help enable successful implementation.
Focusing on data is important, but that may soon become table stakes. Organizations could increasingly benefit from a robust overarching framework that focuses equally on consumers, governance, and the workforce. Key considerations that may differentiate the organizations that are able to implement and scale gen AI from those that aren’t include:
- Establishing effective governance: Organizations may lose gen AI momentum due to complex organizational dynamics. It is important to establish key decision-makers and strategies and then empower teams to test, learn, and build. While organizations can pursue various decentralized or centralized models, center of excellence models have shown greater promise.4 These models tend to centralize expertise, helping to ensure that AI applications are developed and deployed with uniform standards for safety and adherence to emerging regulations. Centers of excellence can foster interdisciplinary collaboration, optimize resources, encourage innovation within a risk-managed framework, ensure ethical considerations in AI deployment, and support scalability and efficient management of vendors.
- Building consumer trust and engagement: Gen AI has the potential to advance diagnostics, clinical decision support, and personalization.5 But first, gen AI may have to gain traction and demonstrate value. Organizations should actively engage consumers to understand their most critical pain points and understand what AI solutions they are willing to use. This can be achieved by gathering direct consumer input and conducting focus groups while iteratively building, testing, and deploying new products. The process should be led with transparency, equity, and collaboration.
- Gaining workforce buy-in: Already overwhelmed with a workforce crisis, health care organizations may want to focus on the needs of their employees and address their fear of this new technology.6 A greater emphasis on workforce literacy and the integration of gen AI as a workforce ally rather than a challenge can help restore workforce trust and alleviate the continued workforce crisis. This approach could help workers focus on the aspects of their roles that help them grow and develop within the changing environment.
- Building solutions for scalability: Today health care companies may encounter hiccups in scaling gen AI solutions to their customers. Standalone gen AI application programming interfaces may face scaling challenges, and no single large language model will likely perform all tasks or use cases. Some of the technical and operational difficulties they could face include implementing robust data pipelines, handling complex documents, integrating custom front-ends, and managing intricate orchestration for prompt handling and vector storage. Designing for scalability upfront by employing robust machine learning operations (MLOps) capabilities can help ensure that machine learning processes are dependable and efficient.
As technology continues to power more health care processes, gen AI likely stands at the forefront of this transformation, with the potential for unprecedented advancements in consumer engagement, patient care, and operational efficiencies. By addressing consumer and workforce considerations alongside the data considerations, health care organizations can pave the way for a future in which generative AI not only augments health care delivery but does so equitably, without bias, in a trustworthy and ethical way, along with a personal touch.
