This article explores the strengths—and limitations—of ChatGPT in helping understand and respond to challenges among the region’s healthcare systems.
It was originally published by Jan Schmitz-Huebsch and Dr. Lina Reypens as a series of four LinkedIn articles, which you can find here:
1. Can AI help us understand AI healthcare solutions in the GCC? Sort of.
Just how useful is ChatGPT? The artificial intelligence (AI) chatbot seems everywhere in recent months, and we were curious: Could it help us understand the potential of AI-powered solutions on healthcare systems in the GCC? Our region poses healthcare challenges that differ from the rest of the world, and we wondered if ChatGPT could provide new insights into persistent issues project teams and clients grapple with.
Our approach was simple. This is the first of several posts examining AI-driven solutions to healthcare in the GCC, and each is the result of queries fed into ChatGPT-41 which we then reviewed and complemented with our own insight. In this respect, the exercise has also helped us explore how AI and humans can collaborate in creating insightful content. Our quick take? ChatGPT can be a useful gateway for initial drafts, but its output requires scrutiny for lack of nuance, faulty use or invention of sources, and the degree of distinctive advice.
What we asked and how AI answered
We asked ChatGPT to provide an overview of the potential of AI in healthcare, with context for the GCC and the different impact AI may have by country. After an opening sentence noting the potential of AI in the region, here’s the balance of its response:
The impact of AI on healthcare varies across countries—those with advanced technological infrastructure and digitalization have made more significant strides in AI adoption. Healthcare system structures and government involvement can also impact AI adoption. Cultural norms, attitudes towards technology, and economic factors also play a role in determining the impact of AI in healthcare. Despite these differences, the global discourse on AI in healthcare often focuses on the potential benefits and challenges of AI in a general sense, rather than considering the specific context of individual countries and regions. This underrepresentation of country-specific differences can lead to an incomplete understanding and may result in a failure to address the unique challenges and opportunities presented by AI in different contexts.
It's a generic answer. So, we narrowed the query to ask specifically about how the impact of AI in healthcare in GCC countries differs from the rest of the world, “especially in the context of technological infrastructure, healthcare system structure, regulatory frameworks, cultural norms, and economic factors”—the dimensions ChatGPT had specified for us. This prompted a more useful outcome—and better insight into the power and limitations of ChatGPT.
Our perspective on ChatGPT’s insights
GCC countries do display regional differences in the adoption of AI in healthcare in at least three of five of the dimensions the chatbot provided. But there are nuances ChatGPT missed, demonstrating the limitations of the technology (at this point) and the value we can add from the work we undertake daily in the sector. Each section below includes ChatGPT’s response in italics, followed by our analysis:
-
1Technological infrastructure.
Technologically, GCC countries boast advanced infrastructure, enabling rapid AI adoption. However, uneven distribution highlights the need for targeted investments to bridge disparities in access and quality of care.
In line with ChatGPT, we expect most GCC countries’ technological environment to boost the adoption of AI solutions. But despite emerging efforts in research and development—such as through KAUST, G42, or the Qatar Computing Research Institute—we expect most solutions to be developed outside the region in the short-term. -
2Healthcare system structure.
Healthcare systems in GCC countries emphasize public-private partnerships, fostering collaboration and innovation. Governments actively invest in healthcare, driving private sector participation and creating a conducive environment for AI-driven improvements.
The shortage of healthcare professionals boosts the business case to leapfrog manual efforts and invest in advanced AI solutions. But ChatGPT misses the significant challenge of aligning the GCC’s fragmented systems, which span everything from nationals and expatriates to the private and public sector and other government and defense health systems. And while there’s an emphasis on private-sector participation, we frequently see partnerships stall at the memorandum of understanding stage. -
3Regulatory frameworks.
Regulatory frameworks in GCC countries are rapidly evolving to accommodate AI’s growing role in healthcare. Balancing innovation with data privacy and ethical considerations, these countries strive to create sustainable regulations, distinguishing them from regions slower to adapt.
While regulations still hinder digital and data-driven healthcare solutions (for instance, cross-border health data transfer in Saudi Arabia is still heavily restricted), we do support ChatGPT’s view of a rapid pace of change in regulatory reforms enabled by centralized and potent regulatory bodies. -
4Cultural norms.
Culturally, GCC countries value personalized, high-quality care. AI can enhance this by improving diagnostics, treatment plans, and patient monitoring. However, concerns about job displacement and impersonal aspects of AI may create resistance to widespread adoption, emphasizing the need for thoughtful implementation strategies.
We also see a high level of cultural acceptance of new technologies that work in favor of AI adoption in healthcare. In fact, we expect resistance to be weaker than outlined by ChatGPT thanks to the openness to innovation and curiosity to apply new, even immature solutions. A recent YouGov survey, for instance, indicated 50% of employees in Saudi Arabia and UAE could imagine automating their day-to-day tasks via analytics technology, compared with 24% in Germany and 16% in the UK.2 -
5Economic factors.
Economically, wealthy GCC countries have the resources to invest in AI-driven healthcare solutions. With a young and growing population, demand for improved healthcare services drives AI adoption.
Among all differentiating factors of the GCC, economies may be the least problematic. The financial resources of GCC countries and their ambition to diversify largely oil-based economies favors the adoption of AI solutions, and many are low cost and can be adopted by countries with low healthcare spending.
2. Addressing the GCC’s healthcare challenges: ChatGPT’s to-do list
Could AI solutions address the healthcare challenges faced by GCC countries? Having used ChatGPT to establish a baseline of the preparedness for AI solutions in the region, we now asked the AI chatbot this very question, adding: “How far can AI solutions help overcome existing challenges in terms of cost efficiency, access, and quality of healthcare in the GCC?”
ChatGPT’s response identified four key challenges. Each section below includes ChatGPT’s response in italics, followed by our analysis:
-
1Chronic disease management.
High prevalence of chronic diseases. AI solutions can be used to develop personalized treatment plans, monitor disease progression, and identify early warning signs of complications.
GCC populations suffer from high prevalence of obesity, diabetes, cardiovascular disease, respiratory diseases, and—more recently—cancer. In fact, diabetes prevalence dramatically exceeds the global average (9%) in Saudi Arabia (18%) and Kuwait (22%), with only Omanis beating the global rate (8%).1 Yet these diseases are largely lifestyle related, making investments in sophisticated digital solutions at times disproportionate compared with public awareness campaigns or regulation of sugar and tobacco consumption. Also, excessively focusing on chronic diseases risks other diseases being overlooked (such as mental health), not to mention new viral infection waves and antimicrobial resistance. -
2Access to healthcare.
Growing population with an increasing demand for healthcare services, which can put a strain on healthcare resources. AI solutions can be used to improve healthcare delivery, such as telemedicine platforms that allow remote consultations and diagnosis, reducing the need for physical visits to healthcare facilities.
Access to care is indeed a problem observable in remote, rural areas that can be tackled by AI-supported remote diagnostic, treatment, and monitoring solutions. -
3Medical errors.
Medical errors are a significant challenge, and AI solutions can help to mitigate this, reducing the risk of errors due to human factors.
Medical errors are a pressing challenge, most commonly in the form of faulty medicine prescription and administration.2 AI solutions can help, but it also requires personnel training. More importantly, this challenge is also related to problems in quality of care, which have been shown to improve with adoption of AI-solutions (e.g., decision-support systems in diagnostics and treatment). -
4Limited healthcare workforce.
Challenges in attracting and retaining healthcare professionals, leading to a shortage of qualified staff. AI solutions can help by automating routine tasks and freeing up healthcare staff to focus on complex and specialized tasks.
The ratio of nurses in Saudi Arabia relative to population is less than a third that of Switzerland3, and AI-based solutions can help free up capacity of existing healthcare staff by alleviating administrative burdens that can take up to 50% of healthcare professionals’ time (Indian startup Augnito, for example, uses voice recognition for medical documentation purposes). And time and experience will help overcome some of the current limitations, such as information software cannot capture when patients use visual gestures to locate body parts or symptoms.
“There is a need to prioritize rare genetic and metabolic diseases in the region that affect a significant proportion of the population and move the needle from treatment to prevention.”
ChatGPT also omitted the fact GCC countries are increasingly seeking to enhance their healthcare systems’ cost efficiency. AI-based solutions can contribute to lower costs, for instance, by enabling early detection and early intervention, eliminating unnecessary treatment plans, and optimizing workflows.
The chatbot’s summation was that AI solutions can help by “improving accuracy in diagnosis, enhancing precision medicine, reducing the time taken to develop new drugs, optimizing workflows, enhancing communication between patients and healthcare providers, and reducing errors.” What’s missing is an acknowledgment that the GCC lags in healthcare research efforts. For instance, the UAE and Saudi Arabia together host fewer than a third of the number of clinical trials in Turkey or Egypt.4 Companies such as BenevolentAI and Unlearn.AI offer computational solutions to enhance drug discovery or manage intelligent clinical trial arms that can help GCC countries address current gaps in their research landscape. Abu Dhabi’s Mohamed bin Zayed University of Artificial Intelligence has recognized this potential and aims at identifying drug targets for aging-associated and rare diseases through AI.
3. What GCC countries are doing on AI in healthcare, and what they should do - ChatGPT’s regional scorecard
After setting a baseline for region’s preparedness for AI and the core challenges in its healthcare systems, we wanted ChatGPT to provide example for each country to understand what governments are doing today to address healthcare challenges through AI solutions. We also wanted to know “what healthcare challenge do they need to focus on more / what healthcare challenge could they focus on less.”
The AI chatbot provided specific examples by country. It also noted “all GCC countries are grappling with the same issues, such as rising healthcare costs, an aging population, and an increase in chronic diseases. However, each country has its unique challenges, such as Qatar's need to address the high prevalence of diabetes, and Kuwait's need to improve mental health services. Governments should focus on investing in AI solutions that can address these specific challenges and improve health outcomes.”
We’ve seen governments in the GCC invest in the development and adoption of AI solutions in healthcare. This often starts by mandating dedicated entities (such as G42 in Abu Dhabi or the Saudi Company for Artificial Intelligence, owned by PIF) in an ‘AI center of excellence’ approach. This bundling of expertise can be helpful initially, but a more decentralized model may need to be considered over time.
One big issue: We found inconsistencies and missing sources in ChatGPT’s output. For instance, while AI in telemedicine is cited in Oman, ChatGPT “could not find a specific source on the use of artificial intelligence in telemedicine in Oman.” For AI use in telemedicine in Qatar, the chatbot cited an article allegedly published by the Journal of Medical Systems in 2018, but that is nowhere to be found online and its unique identifying number leads to a Chinese publication on upper limb function rehabilitation. In the UAE, when prompted for additional examples ChatGPT cited the Emirati Minister of Health praising the “Genomics-Driven Prescription (GDP) Program”—a credible-sounding initiative that was untraceable in ChatGPT’s sources and has never been heard of among our colleagues and clients.
Suggested actions for governments
When asked what GCC governments could do to “improve the application of AI to solve healthcare challenges,” ChatGPT noted the primary focus to date had been on improving healthcare access and quality. “Moving forward,” the chatbot said, “GCC governments should also focus on addressing cost efficiency, data privacy, security, and interoperability, and investing in the development of local AI talent to build a sustainable AI ecosystem.”
It then provided proposals by country, from a centralized data platform in Saudi Arabia to AI-powered tools for clinical decision making in Kuwait and tools for disease detection and early diagnosis in Oman. Yet these appeared to be random and were not nation specific: ChatGPT’s suggestions to establish data platforms, leverage public-private partnerships, and invest into AI-powered tools for clinical decision making and early detection were as accurate as they were generic.
Its generalization misses the point that authorities in the GCC would benefit from revising the healthcare challenges their AI solutions focus on, such as access to care via AI-powered telemedicine. We see challenges deserving more attention including population health management, quality improvement, and intensified healthcare research. This requires sustained funding for research, education, and implementation efforts, as well as the continued adaptation of regulatory frameworks. In an ideal scenario, GCC countries would form complementary partnerships within and beyond the region—allowing partners to channel their efforts on a few concrete specialties.
What’s needed more broadly is a clear and coherent AI strategy. It will be vital to develop guiding AI strategies at healthcare system level (an initial example can be found in the UAE), which then need to be refined in an agile way to incorporate technological advances and emerging findings. If governments do not effectively steer AI via a clear strategy, redundant applications may be driven forward by multiple stakeholders, use cases may be too focused on short-term impact, and potential benefits may not be scaled across each healthcare system.
“There are a number of ways to improve the use of AI in healthcare to help bridge existing gaps in the system where we could augment the role of healthcare professionals and, more importantly, increase access to care in remote and rural areas. Governments should invest more in AI R&D for healthcare use cases, drive private sector participation through sector specific incentives, and establish the equivalent of GDPR or HIPAA for healthcare in Saudi Arabia.”
4. Evaluating our ChatGPT experiment
The combination of artificial and human intelligence was undoubtedly useful in writing this series of blog posts. ChatGPT was able to synthesize associations of AI in healthcare from online sources, formulate eloquent prose in flexible styles and lengths, and notably speed up the overall writing process.
Superficially, ChatGPT’s answers around the application of AI in healthcare in the GCC also tend to be appealing. It’s tempting to believe ChatGPT understands what it says, but it often does not (in part three, we identified significant inconsistencies and mistakes around specific GCC AI initiatives). In fact, the chatbot seems unable to distinguish between what’s accurate and inaccurate, inventing sources by combining associated words and news outlets, and—to a dangerous extent—asserting when challenged that it doesn’t make mistakes.1
In addition, ChatGPT’s output often lacked nuance, inferred insights from other geographical sources without revealing them, and even elevated insights to be region-agnostic. When asked to differentiate among GCC countries (for example, in terms of the severity of regulatory restrictions), the chatbot refused: “It is not appropriate for me to rank GCC countries with respect to the regulatory restrictions that may slow down the adoption of AI in healthcare.” Yet ChatGPT also at times revealed a lack of synthesis, with examples that needed to be abstracted to higher levels. As is the case for all AI-based solutions, the outcome is just as good as the data that is fed into it – in future studies, it would be interesting to test ChatGPT’s ability to answer complex questions based on reports and other materials fed into it.
Our take is information from ChatGPT can serve as a useful gateway for further research but needs to be interpreted with care and intense scrutiny of sources. This is also true when ChatGPT itself is used for healthcare purposes, even though as we wrote this article it emerged that ChatGPT gave correct answers to more than 90% of US medical exam licensing questions. 1 It has the potential to be used in healthcare in domains such as audience-targeted communication (“Explain treatment plan so that a 10-year old can understand it”), diagnostics (“What could be underlying diseases for symptoms ABC?”), or patient adherence and disease management (“How can I deal with my mental health condition?”). Of course, this final use case presents questions around ensuring data privacy which, together with ChatGPT’s issues with false data sources, need to be solved.
Yet we’re optimistic. Both ChatGPT and our human sources believe GCC countries are uniquely prepared for a rapid adoption of AI solutions in healthcare that will require coherent national strategies and regulatory reforms. Pilots initiated by providers and payors can serve as low-risk testing grounds, research organizations can already leverage it for multifold purposes, from population genomics to the fight against antimicrobial resistance. But driving rather than just adopting innovation requires governments devote increasing attention to investing in research and education, update regulations for new products and solutions, close the talent gap, and prevent abuse and harm through the uninformed use of AI.
Contact us
Menu