"Check out our podcast episode on the future of AI in medicine!"
"Check out our latest podcast episode on the importance of future of calculus" It is now on Spotify!
"Check out our podcast episode on the future of AI in medicine!"
Can AI Replace Human Doctors in the Next 50 Years? By Ellie Lau
‘Hi there, how can I help you today?’ It’s a question we saw from an AI chatbox, but what if, in the future, it came from an AI doctor diagnosing and treating patients? With the rapid development of technology today, people are becoming increasingly reliant on AI. Recently, reports have shown that AI in hospitals is diagnosing conditions more accurately than doctors, increasing public attention to the question ‘Can AI replace human doctors?’. It is undeniable that diagnostics based on a large amount of data will definitely exceed the years of experience accumulated by an individual doctor. However, the question remains: By 2075, will AI be advanced enough to fully replace doctors?
One of AI’s most popular applications in medicine is in diagnostics. I would like to bring up a real-life example that highlights how AI is essential to medical development. The Royal Free Hospital in London in collaboration with DeepMind Health, developed an AI app called Streams, designed to help doctors to detect acute kidney injury (AKI). This system continuously analyses patients' electronic health records in real-time, recognising early warning signs. This helps to alert the doctors to take intervention measures before the condition worsens. In one case, a patient was in danger due to a sudden deterioration in their condition but Streams detected signs of acute kidney injury early and sent an urgent alert to the doctors. This prompt notification enabled the doctors to take immediate action, preventing the patient's condition from worsening. It is shown that the AI system has significantly improved the hospital's response time and saved both time and lives.
Some argue that AI could address the global doctor shortage. AI can process vast amounts of medical data instantly, reducing the workload for doctors and could reduce healthcare costs. Becoming a doctor is a very long and rigorous journey. In the UK for example, after a long process of UCAS applications with many medical interviews and the UCAT test, it takes at least 5 years more to train a junior doctor while sometimes 6. According to the British Medical Association, the UK has a low proportion of doctors relative to the population. Only 2.9 doctors per 1,000 people compared to 4.3 in Germany. This has left the UK government with no choice but to start implementing policies aimed at doubling the number of medical school training places. Nevertheless, AI could further bridge this gap by handling some routine tasks so that doctors can focus on more complex cases and direct patient care.
AI can also be involved in telehealth, providing patients, especially those in underserved or rural areas, with access to healthcare services that might otherwise be unavailable. For instance, Babylon Health uses virtual consultations and symptom-checking tools to diagnose conditions and provide medical advice. All you need to do is answer a few questions and you can then see your 5-year disease risk across 20 key diseases. It analyses patient’s symptoms and medical histories to offer some personalised advice or direct patients to the appropriate level of care. This is particularly beneficial to rural areas where access to medical professionals may be limited, or for patients who cannot easily travel to clinics or hospitals. In this way, patients who are unable to visit a doctor in person can still receive timely advice and medical attention, enabling them to manage their health remotely. It can be argued that telehealth can ease the burden on the workload of medical staff, hospitals etc. And this can also be used during a pandemic like COVID-19 in 2021 where people avoid going into the hospital area since there is a high risk of being exposed to the virus.
There is another impactful example that demonstrates AI's importance in Medicine. Cancer, one of the most complicated problems in the world, despite the development in medicine, there is still no guaranteed cure. However, AI is starting to make a real difference in early detection and treatment. An AI model created by researchers at MIT, Sybile can analyse scans and spot early signs of lung cancer. Even more impressive is that it is up to 90% accurate, giving doctors a powerful tool to detect cancer earlier when treatment is the most effective. As we all know, before starting a cancer treatment, doctors have to spend hours carefully marking organs on scans to ensure that radiation targets the cancer cells while avoiding damaging the other healthy tissue. This process can take up to three hours per patient, but AI-powered tools like OSAIRIS, developed for NHS England, are changing that. With AI’s help, oncologists can now plan radiotherapy 2.5 times faster, reducing the waiting times for patients and making treatment more efficient.
Despite these advantages, AI has significant limitations that prevent it from fully replacing human doctors. Medicine is not just about diagnosis and data, it also involves factors beyond data such as communication and empathy with patients. Trust is a big issue. Many people still feel more comfortable talking to a human doctor rather than relying on a machine. Doctors don’t just diagnose illnesses and prescribe medicine, they will comfort, reassure, and build relationships with their patients. Imagine an AI telling a patient, ‘Don’t worry, I'm here to take care of you to the best of my abilities, and my professional database will try its best to analyse all the data from those similar cases.’ Currently, AI cannot achieve this level of interpersonal communication and emotional understanding, as it lacks of true empathy and human emotional intelligence. In the medical field, ethics and responsibility are core issues. AI systems may lack understanding of ethical principles when making decisions, such as in handling patient privacy or making life-or-death decisions. Additionally, if an AI makes a mistake on an incorrect diagnosis or recommends an ineffective treatment who will be the one that is responsible? Unlike human doctors, AI cannot be held accountable for its decisions, making liability a complex issue. Therefore, AI cannot fully replicate the richness of human experience that is central to qualitative research, communication skills, and ethical judgment of doctors. AI is innovative but it is still unable to offer the understanding and emotional support that a real person can provide.
While it is still controversial whether AI can replace doctors or not, no matter how advanced AI becomes, it still has one major limitation which is it cannot think for itself. AI is great at recognising patterns and processing huge amounts of data, but it does not have true creativity or independent decision-making skills and the ability to adapt beyond its programming. Medicine isn’t just about following a set of instructions, doctors often have to think outside the box, especially when dealing with rare diseases or unusual symptoms that do not fit standard diagnostic patterns. AI can only work with the existing data it has been trained on, which means if that data is biased, incomplete, or outdated, it can make mistakes. For instance, if an AI system is trained primarily on data from Western hospitals, it might not work as well for patients from different ethnic backgrounds, leading to diagnostic inaccuracies and unequal treatment. Similarly, some rare conditions may have limited recorded cases, meaning AI may lack sufficient information to make an accurate assessment. In contrast, human doctors can recognise patterns even in unfamiliar cases, drawing from their past experiences and medical judgment.
AI also faces challenges regarding its vulnerability to cybersecurity threats. There are always risks of sharing your personal information on the internet, and the same concept applies to the potential risks from cyberattacks or data theft by relying on AI systems. As I have mentioned earlier, AI systems rely on vast amount of sensitive patient data such as medical histories, genetic information that are extremely vulnerable to cybercriminals. In June 2024, the system of NHS Synnovis, which is a pathology testing organisation was attacked and almost 400 GB of patient private information was stolen by the Qilin cybercrime group. It shows that in the future, hackers may target AI systems for data theft and sell all the patients' privacy onto the black market. Beyond that, data manipulation is also a main issue, hackers may also alter patient data such as changing the medical histories, diagnoses or test results. AI systems make critical healthcare decisions based on the data they process, so any changes could result in inaccurate treatment plans, misdiagnoses, or even patient deaths. Despite stealing or altering data, another cybersecurity threat lies in hijacking. If cybercriminals take control of the AI systems themselves, it will highly disrupt hospital operations, leading to delays in diagnosis or treatment.
Last but not least, as AI rapidly replaces human labour, it is bound to impact certain job areas. Dr. Radvanyi, the President and Scientific Director of the Ontario Institute for Cancer Research had mentioned that, “Doctors will be the ones that dispense the therapy, not the computer. We need to approach AI as a powerful tool to help us make those decisions.” AI may never replace human expertise, but it’s proving to be an invaluable partner in the fight against cancer. AI-driven companies will likely dominate many industries in the future. The same applies to the healthcare industry, where AI-based check-ups, imaging reports, and pathology reports are expected to become more common, reshaping the role of medical professionals. But in my opinion, AI is never a replacement. I do believe the most effective approach would be AI and human doctors working collectively, like AI could handle data-based tasks, allowing doctors to focus on patient care and human interaction. I believe the future of medicine will be one where AI and doctors work hand in hand to create a healthcare system that is more efficient, precise and patient-centred. Only time will tell how this unfolds.
Reference List
NHS England(2024) Culture and leadership, Healthcare inequalities, Long Term Plan, Safe staffing, Workforce Available from: https://www.england.nhs.uk/long-read/nhs-long-term-workforce-plan-2/#:~:text=Train%20%E2%80%93%20Grow%20the%20workforce&text=This%20Plan%20sets%20out%20the,and%20help%20address%20geographical%20inequity (Accessed 17th March 2025)
NHS Royal Free London (2018) Royal Free London publishes audit into Streams app Available from:
https://www.royalfree.nhs.uk/news/royal-free-london-publishes-audit-streams-app (Accessed 16th March 2025)
Intuitive (n.d.) Da Vinci robotic surgery Available from: https://www.intuitive.com/en-us/patients/da-vinci-robotic-surgery/about-the-systems (Accessed 18th March 2025)
Alexandra F. Gibson Alexander Beattie (Feb 2024) More or less than human? Evaluating the role of AI-as-participant in online qualitative research Available from: https://www.tandfonline.com/doi/full/10.1080/14780887.2024.2311427 (Accessed 18th March 2025)
Bennett Institute (2024) The-case-for-an-AI-driven-personalised-prevention-platform Available from: https://www.bennettinstitute.cam.ac.uk/wp-content/uploads/2024/03/The-case-for-an-AI-driven-personalised-prevention-platform.pdf (Accessed 18th March 2025)
Babylon Healthcare (n.d.) Babylon’s AI Healthcheck: helping people better understand their
physical and mental health Available from: https://assets.babylonhealth.com/redesign/Babylon%E2%80%99s-AI-Healthcheck-press-release2.pdf (Accessed 28th March 2025)
healthsnap.io (n.d.) Available from: https://healthsnap.io/ai-in-remote-patient-monitoring-the-top-4-use-cases-in-2024/ (Accessed 29th March 2025)
Sebastian Klovig Skelton (Jun 2024) Qilin ransomware gang publishes stolen NHS data online Available from:
https://www.computerweekly.com/news/366589583/Qilin-ransomware-gang-publishes-stolen-NHS-data-online (Accessed 29th March 2025)
Union for International Cancer Control (UICC) (Jun 2024) Podcast "Let's Talk Cancer" – The future of oncology: new technologies and tailored treatment Available from: https://www.uicc.org/news-and-updates/news/podcast-lets-talk-cancer-future-oncology-new-technologies-and-tailored (Accessed 1st April 2025)
Union for International Cancer Control (UICC) (Feb 2024) How is Artificial Intelligence (AI) shaping the future of cancer control? Available from: https://www.uicc.org/news-and-updates/news/how-artificial-intelligence-ai-shaping-future-cancer-control (Accessed 1st April 2025)