Why AI Still Falls Short in Solving Healthcare’s Hardest Problems

Shahid Akhter, Consulting Editor, FEHealthcare, talks to Ritu Mehrotra, Founder & CEO, Shunya Labs & United We Care, to understand how AI is reshaping healthcare—where it falls short today and what unique innovations her teams are building to address gaps in access and clinician support.

 

Over the last few years, AI has advanced rapidly, but where do you think it still falls short in healthcare?

AI has been around for a very long time, but it received a huge push after the emergence of OpenAI. In the last four to five years, trillions of dollars have gone into AI advancement, and many companies have entered the AI space due to this surge. However, in healthcare, AI is still not deep enough. The adoption and meaningful impact remain limited compared to other industries, which highlights that despite the progress, healthcare still lacks the depth of AI integration needed to solve critical problems.

 

 Access and affordability seem to be the two biggest issues you mention—how critical are they to solving global healthcare challenges?

Access and affordability are absolutely critical in solving global healthcare challenges. To begin with, more than 25% of the world does not have access to healthcare. For example, 70% of rural America faces a shortage of clinicians, and India’s remote areas suffer from the same issue. In developing or underdeveloped regions such as Africa, the problem is even worse, while even in developed countries like Scandinavia, clinician shortages exist. 

Affordability is equally pressing, as 800 million people worldwide spend over 10% of their income on healthcare expenses. Many people are pushed into poverty due to the lack of insurance, and several countries do not have strong insurance ecosystems. By 2030, the global clinical shortage will be so severe that 30–40% of people will not have access to the clinical ecosystem, which makes it imperative for AI to step in and play a critical role in bridging both access and affordability gaps.

 

 There has been debate globally about whether humans should be “in the loop” with AI. What’s your perspective on that for healthcare?

In the last five to six years, significant advancements have been made in diagnostics, prescriptions, and bridging healthcare gaps. However, the problem remains that AI is not yet accurate enough. This has sparked a global debate on whether humans should be in the loop, on the loop, or over the loop. The reality is that healthcare cannot depend solely on AI; it requires a balance with human oversight. Human intelligence and AI are not substitutes but complements—working hand in hand to ensure accuracy, trust, and better healthcare outcomes.

Healthcare has one of the lowest levels of AI adoption, primarily due to accuracy challenges. Clinicians spend nearly 40% of their time on paperwork, and in the US, where 93% of people are insured, this burden is especially heavy. For AI to be truly effective in this domain, it must go beyond generic language models—it needs to accurately interpret medical terminology, context, and clinician intent. A healthcare AI system must therefore be designed with high precision, strong accuracy, and advanced intent detection to meaningfully reduce clinician burden and build trust among medical professionals.

 

 You mentioned precision, reasoning, and intent detection as crucial gaps. Can you explain why they are so important in healthcare compared to other industries?

Precision, reasoning, and intent detection are extremely important in healthcare because, unlike other industries, even small errors can have life-or-death consequences. For example, while platforms like Meta are able to detect user intent and show ads accordingly, in healthcare, intent must be captured at the deepest level of detail. Similarly, reasoning engines are essential. In cancer diagnoses, doctors often draw on X-rays to highlight specific lobes and problem areas. While OCR can read text, it cannot interpret this context without a strong reasoning engine. AI must both understand language deeply and reason with context to be truly useful in healthcare. Although AI has advanced from a basic to a more complex reasoning level, hallucinations remain a major challenge, making human oversight essential. 

One area where AI has excelled is robotic surgery, as machines can perform precision-based tasks like neurosurgery or shunt placement better than humans. The future potential also lies in post-surgical care, where AI can handle routine tasks such as monitoring vitals, pharmacogenomics, or surgical results to reduce the doctor's workload. Despite these advancements, hallucinations are still common, though patents and solutions are being developed. The comparison to the dot-com boom is apt, as massive infrastructure investment will be required, but adoption and progress are moving much faster. My personal journey as a cancer survivor further reinforces the urgent need for AI to improve access and reduce clinician shortages.

 

 What unique approach are United We Care and Shunya Labs taking to solve these issues in a way others haven’t?

At United We Care and Shunya Labs, we are uniquely addressing these challenges by focusing on solving the global clinical shortage and affordability issues in healthcare. Unlike generic AI systems, our approach is to go deeper into healthcare-specific AI development. 

We began by creating an Automated Speech Recognition (ASR) system that broke seven world records. This ASR system works on high-entropy data, uses exponential pruning, and delivers extremely high accuracy across 30 Indic languages, outperforming tech giants like Google, Nvidia, and Meta. What makes it unique is that the model is small enough to run on a CPU or mobile phone, avoiding the need for costly GPUs, which makes it more accessible to doctors worldwide. 

In addition to speech recognition, Shunya Labs built a reasoning engine capable of handling critical differences, such as distinguishing between a patient being allergic or not allergic to penicillin—a difference that could save a life. This service model was first launched for mental health in Australia, where the system streamlines the patient intake process, records doctor consultations, automatically generates structured notes, and routes them for insurance and prescription purposes. It also acts as an engagement tool by reminding patients of medication schedules and follow-up visits and answering questions between consultations through AI. Together, Shunya Labs and United We Care are building a complete, affordable, and accessible AI-powered healthcare ecosystem that is uniquely designed for real-world deployment.