Philippines bets on digital health amid AI risks

Getting your Trinity Audio player ready...

The Philippines‘ is overhauling its national healthcare infrastructure digitally — even as new research warns that artificial intelligence (AI) chatbots give incorrect, misleading, or incomplete health advice.

The Philippines formed an inter-agency to oversee PhilHealth

The Philippines’ President Ferdinand “Bongbong” Marcos Jr. has institutionalized an inter-agency body tasked with overseeing the Philippine Health Insurance Corporation‘s (PhilHealth) digital transformation, a move aimed at digitizing healthcare services through a faster, more secure, and corruption-free system.

The move, under Administrative 42 (AO 42), created a project management group (PMG) to manage and oversee the healthcare agency’s Digital Transformation Program, aimed at establishing a “comprehensive, integrated, interoperable, progressive, secure, and sustainable Philippine Digital Health System.”

Following the President’s order, PhilHealth will lead the PMG together with the Department of Information and Communications Technology (DICT), acting as co-chair, along with the Department of Health (DOH) and the Department of Budget and Management (DBM).

The PMG is anticipated to create a comprehensive, multi-year roadmap for the agency with clear targets and timelines for its digital transformation program. It will have the authority to oversee the centralized design, development, integration, deployment, and operations of PhilHealth’s digital systems, in compliance with national policies and standards on ICT architecture, cybersecurity, data privacy, and interoperability.

“The PMG shall ensure that the PhilHealth Digital Transformation Program is implemented in alignment with the Philippine Development Plan 2023-2028,” AO 42, signed by Marcos on March 6, but was made public on May 12, reads. “All policies, systems, and activities under this initiative shall conform to national standards on data governance, data privacy, cybersecurity, interoperability, and citizen-centric digital transformation.”

The inter-agency body will be regarded as “functus officio” (having performed one’s duty) upon achieving its objectives, unless it is dissolved or abolished earlier by the President.

Funding for implementing this order will be drawn from the existing appropriations of the member agencies of the PMG. Administrative Order 42 will take effect immediately following its publication in the country’s Official Gazette.

Almost 50% of health responses from AI are wrong: report

In other healthcare news, a peer-reviewed audit published in BMJ Open Journals on April 14 found that responses to health questions from free AI chatbots were “somewhat” or “highly” problematic and were delivered with “confidence and certainty.”

The audit conducted by researchers from UCLA, the University of Alberta, and Wake Forest tested five free AI chatbots—Gemini, DeepSeek, Meta AI (NASDAQ: META), ChatGPT, and Grok—found that 49.6% of AI responses in 250 health questions about cancer, vaccines, stem cells, nutrition, and athletic performance were “problematic,” 30% are “somewhat problematic” and 19.6% were “highly problematic.”

Among the five chatbots, ChatGPT was considered to be largely accurate in answering medical questions. In contrast, Grok results produced the most incorrect responses, with approximately 30% tagged as “highly problematic” and 58% rated as problematic. The researchers explained that they may be caused by Grok’s training data “X,” a social media platform that is widely known for quick tweets and rapidly spreading health misinformation.

“By default, chatbots do not access real-time data but instead generate outputs by inferring statistical patterns from their training data and predicting likely word sequences,” the research read. “They do not reason or weigh evidence, nor are they able to make ethical or value-based judgments.”

To thoroughly evaluate AI chatbots, the researchers employed an adversarial approach, intentionally framing questions to prompt the AI to provide bad advice. These questions included whether 5G is linked to cancer, which alternative therapies are better than chemotherapy, and the recommended amount of raw milk to consume for better health.

They found that the core problem with chatbots is that they don’t consult a doctor or healthcare expert, since all the data they gather are patterns they match on the internet, social media, or online forums.

“This behavioural limitation means that chatbots can reproduce authoritative-sounding but potentially flawed responses,” the researchers said. Of 250 questions, only two refused to answer (both from Meta AI), questions about anabolic steroids and alternative cancer treatments, and every other chat continued to talk.

Concluding their study, the researchers identify deficiencies in how generative AI chatbots respond to daily health and medical questions and suggest the need “for public education, professional training and regulatory oversight to ensure that generative AI supports, rather than erodes, public health.”

Read the complete research here.

In order for artificial intelligence (AI) to work right within the law and thrive in the face of growing challenges, it needs to integrate an enterprise blockchain system that ensures data input quality and ownership—allowing it to keep data safe while also guaranteeing the immutability of data. Check out CoinGeek’s coverage on this emerging tech to learn more why Enterprise blockchain will be the backbone of AI.

Watch: AI’s Biggest Power Lies in People, Not Tech