Webit 2026: AI in healthcare: the hope, the hype and the reality - Prof. Shafi Ahmed

This talk focuses on the intersection of AI and healthcare, highlighting the tremendous potential and ethical challenges presented by AI technologies in medical diagnostics, patient care, and surgical procedures. Professor Shafi Ahmed discusses how AI can enhance human capabilities, improve decision-making, and increase healthcare access, while also stressing the importance of trust, ethics, and explainability in AI application. He emphasizes the need for responsible implementation of AI technologies, particularly in areas like radiology and dermatology, where AI has shown promising results. Furthermore, he explores how AI can facilitate medical training and improve surgical practices through innovations such as virtual reality and AI-assisted workflows. The session ultimately calls for a balanced approach to integrating AI into healthcare, ensuring it amplifies human effort rather than replacing it.

After a day of conversations about infrastructure, enterprise systems, [music] banking, robotics, creator attention, regulation, and AI native organizations, we now turn to perhaps the most human frontier of all, healthcare. Because healthcare is where [music] intelligence, trust, ethics, data, human skill, and life outcomes collide. AI and medicine carries extraordinary promise. From diagnostics and surgery to precision treatment, medical training and patient access, it can augment human capability, support

doctors, improve decisions, and help bring better care to more people. But healthcare is also where the gap between hype and reality matters most. The stakes are too high for shallow promises. AI in healthcare must be trusted, ethical, explainable, responsible, and deeply human centered. This session explores both the hope and the hard questions. What AI can truly change in medicine where implementation remains difficult and how we ensure

that technology amplifies humanity rather than replacing it. Please welcome Professor Shafi Ahmed, senior partner at Purpose Tech Capital for AI in healthcare, the hope, the hype, and the reality. Thank you so much. It's great to be here back in Bulgaria. I think this is my 10th time I've been invited back over the last decade. Uh PLM and I met a decade ago on Webbit and I've been

always amazed by the amazing innovation in this part of the world. The weight of now doubt is not a pleasant condition but certainty is an absurd one. There is a particular kind of discomfort that comes from standing in the middle of a transformation that you cannot yet fully see. It is not fear exactly. It's not excitement. It contains elements of both. It's a sensation of knowing that

the ground beneath you is moving. That the map you have been used to navigate it was drawn from a different terrain. That's the opening prologue of my new book on on AI. It's called intelligent uh the evolution of AI transforming health care. I thought I'd start using that prologue and the the quote above you now which is the punctuation of everything. If you look through evolution, one

famous person of course that we've come across in that theory uh talking about our theory of evolution was Charles Darwin. Charles Darwin was an evolution biologist based in the UK. He traveled to the Galapagus Islands on the HMS Beagle where he looked at slow change and incremental change. He described a theory that slow to adapt, slow to change over millions of years, small changes in tortoise shells,

small changes in the shape of birds speak. It's a good theory. In 1970 that theory was challenged by a person called Steven Gourd, a professor from Harvard. He described a different kind of um evolution which he described as punctured equilibrium. He said society moves slowly as described by Darwin that suddenly we change our trajectory forever. We move we forced to move. Each of those he called a

punctuation. We know those now going back through history. The advent of fire. Prior to fire we are Neanderthal living in the dark. The light bulb. Electricity. The printing press by Gutenberg. The motor car. Before the motor car, humans only traversed about 10 miles outside their existence. They didn't know the world outside that. And of course, the computer by Charles Babage in the 1950s. Another punctuation and perhaps

the ultimate punctuation is of course launch of LLM's chat GPT on 30th November 2022. It changed our civilization forever. Suddenly AI was available. It was acceptable. It was usable. It wasn't technical. It wasn't aspirational. It wasn't mild in science. Everyone in the world could use an AI platform all of a sudden. There is a weight of now. When I wrote the book, I felt that there's a

weight on all of our shoulders. The pressure doubt is important. Doubt is essential in AI. We don't know where AI is heading towards. The world of super intelligence, the singularity, can we control it? Can we contain it? Healthcare is one of those areas and verticals. We have to contain it. We have to control it because the the the benefits are amazing. the risks of patient outcomes. Therefore,

we have to much more careful how we deploy. I remember being at the Oxford Union two months ago. Oxford Union is the oldest debate site in the world. And the debate was this. The motion was will AGI artificial general intelligence be humanity's last great invention. What a question for humanity. Is this the end of a we we are? Is it going to define rest of civilization forever?

I voted against it. Against the motion. We won the motion. We won the debate. First time the debated AI at the Oxford Union for about 5 years. It was consequential. The consequences are enormous for healthcare. The debate has to be had. Do we know about the about the future? We don't. The uncertainty is actually quite We know of course the triangle of the pyramid [clears throat] of

going towards this from large language models to agentic AI to super intelligence and AGI. Of course, this is where we're If we understand the concepts and the direction of flow, it allows us to manage healthcare in a different way. Alan Turing was of course a famous cryptographer based in the UK. He he deciphered the Enigma machine that Nazis were using to send messages. As you may remember,

I was at the Imperial War Museum just last week looking at the Nigma machine. Can machines think? Can humans think? Do we work together to provide better outcomes for health care? This was the original machine that he created. Huge machine to decipher the enigma. But of course the first thing to think about is the radiologist. Healthcare is being disrupted by technology particularly of course radiology. This was

a famous quote by Jeffrey Hinton who's the godfather of um of uh I guess the AI in healthcare. He said in 2016 that people should stop training radiologists within five years. In five years, Mayo Clinic America appointed 40% more radiologists. They didn't listen to Jeffon. We need more people, not less. Jeffington made one fault mistake. Healthcare isn't the other vertical. To put AI into a system, it's

about change management. It's workflow management, making more efficient, not add barriers of friction to people's management of disease. Of course, imagenet by faith ling was that first real platform and technology that allowed us to use images in a way that we use it now. For example, imagenet came first. Then radiology came. We're seeing now the deploying of so many solutions for msk for CT scans for brains

uh for chest x-rays and for TV. The only platform that's really democratized health care in radiology is TB diagnosis. Imagine a rural setting in India where you don't have radiologists or radiographers. You can take a chest X-ray, AI would tell you if it's TB or not. That person can travel then hundreds of miles to get treatment. It's the only solution so far that's shown global outcomes. Here's

a problem. Of course there are thousand radiology solutions in America with FDA approval. So 2% are being used. I wrote I write in the book the reasons why we haven't deployed these technologies over the last five years. The friction the change management the difficulty in adapting to new changes within healthcare but they promised much. This was Jeffrey Hinton only more recently revisiting the whole statement. He responded

to Twitter feed and he said, "Of course I was wrong at the time." And this is the problem. AI has had false promises. It's overpromised and underdelled. In some way, that's good because it allows us to think about containment and control of AI so that healthcare can be managed in a way that's safe and responsible based on ethics, governance, and of course, security. If you look at

the next part of my book, I call it what the eye can see. Our eyes are limited. They're biased. With the brain, of course, we only see certain things. We only get trained in certain ways. This is my own eye. I had it magnified over a thousand times. This what it looks like reality. It's flawed. We make inferences based on our knowledge base. Can we do better

with AI? We can. This is a colonoscopy. This is an image. This is a video being taken. We're looking for a small polip which might become cancer over the course of few years. We can't see at the moment as with our eyes. The AI has picked up a lesion's focused it says here you are. I'm assisting you and now you can see the lesion there. It's taken

a while for us to see it but it's there. Take that away you you prevent a cancer in five years time. It's aiding us. Dermatology is a real good one of course with AI. Lots of platforms out there being used to diagnose skin lesions. The first company in the world has got class 3 EU approval for skin That to me is incredible. Last year, what does that

mean? Derm, a platform by skin analytics, can take a picture of your skin lesion. Take a picture. If that image says it's cancer with AI, it's a cancer. It says no cancer, you can walk away. There's no human in the loop. For the first time in history, there's no it's autonomous. There's no doctor saying I need to see this and advise you. That for me is absolutely

monumental in our history of AI. But they based on 10 years of research and data before they got approval. Can you imagine your doctor will certainly see you now? Education is changing. We know that of course AI has been used in AI in in medical learning. And what we're seeing now is a whole way of of teaching and learning. I've been using technology for many years in

virtual reality, Google Glass, smart glasses, metaverse, and we've built all those on AI platforms. Now we can go into a virtual medical school. You can have a virtual Oscar training. AI will mark you in the virtual world and give you a mark. You could pass entire medical school in a virtual world theoretically, but we're creating these platforms to see what we can push boundaries of innovation technology.

A lot of people mention agentic AI and that's kind of where we are now. If you look at the curve of where we are, agentic AI is the middle and physical AI is coming uh quite rapidly. The inverse law suggests that we use technology and improve health care where it's less needed. Where it's more needed, we don't deploy as well. The inverse care law is something that's

stood the test of time for over 30 >> Hi, what task can I help you with today? >> IQ. I'd like you to reach out to patients that haven't done their smear test yet and explain to them why they should do it. >> Would you like me to give you a short draft of this task? >> Yes, please. >> Here's a draft for your clinical task. >>

So, this is one of our comes called quadriia. We built genentic AI so that we can literally say to the AI go to the database and health care in our health record find those women between 18 and 25 find out who has had a cervical screen. If they haven't can you phone them every day until they respond. When they respond you give an appointment and let us

know how you get on. That's the building we do AI the back end building platforms that help us manage the workload. I'm a surgeon. I've been a cancer surgeon for over 30 years now. And I've seen much change in my 30 years. We've we've gone from analog surgery to digital surgery, a new kind of concept for where we are today. Abel Kasim alas was a uh surgeon

based in Ander Lucia around a th00and AD. And he wrote a book called the um Alas which became the standard textbook for surgery for 500 years. Yes. Not one year or two years, for half a millennium. There's no real innovation for a long period of time. his punctuation was the book. After that, there was no innovation for a long time. And so, our world is changing. This

world that I live in now, it's a futuristic world of healthcare, innovation, surgical practices, 3D printing, uh, AI platforms, navigation tools. Super exciting. This is our reality. This is what I live in now. Search navigation, AI tools allowing you to navigate through an operation, tell you where the structures are. robots helping us navigate and do small and small operations much more precisely, much more accurately, using XRVR

for training and allowing us to assess performance using AI Super exciting, allowing us to reshape the way we practice surgery. The operating theater has changed. This is a company that I work with called Vitio. test it on their board. They place cameras in the O like the Amazon Go shops where you can see everything's going on. Analyzes everything that's happening, assesses performance, looking efficiency, how we navigate.

This is what it's capturing. It takes that information, inserts into a dashboard allowing to make those changes and understand the operation, the complications, what's going on, how the team are working to improve the experience, to improve the outcomes. So we're now deploying that on a global level. So AI in the O. Of course, listening is what we do most. We're listening machines as humans with patience and

human interaction. We listen. It's one of our cardinal um I guess uh things that we have to help our patients. Over the last two years, 1.8 billion has been invested in ambient AI scribes. We use them all the time now. Heidi, tandep, all the others now are being used routinely to take a history, insert that conversation into health records. AI scribes is the lowest hanging fruit in

the path of AI. There's no risk. You take a story, take a history, take conversation, takes that, captures it and formats it into a health record. The others are much tougher. That's why scribes are so popular at the moment and disseminating rapid. I was in Toronto last month. 10% of the rural population of uh of Toronto GPS are now deploying it tand uh using AI scribes to

manage their workload. We live in a world of information. Listening machine is one thing. We have other listening machines. Data being collected through our rings, through our smartwatches. We're seeing this being collected on a regular basis. Now that data and AI insights give us world longevity and the world of healthcare and of course what we've become is a quantified self. We're now becoming a bunch of numbers.

Can we use those data points to improve our health to prevent illness to prevent cancer to make us much healthier? Well, I think we can as we go forward. Last two slides is around our code of life. Bill Clinton sat on the lawn of the White House in 2000 when the human genome had just been sequenced. He made the statement, "We have learned the language in which

God created life, unraveling DNA. That's led to drug discovery. It's led to protein folding. Our full team of course understood protein folding which led to drug discovery at scale. The third drug now has been produced on AI. It normally cost about a billion dollars over 10 years. This has cost a few million over a few years. Class three drug in phase three trials. And if you go

back to Alph Go, there was one thing that they mentioned when they produced this platform to beat the world champion of Go Lee Cidle. Move 37 was the thing they said. The computer played this game which 100 millions of ways playing it. Found a new way of playing the game. It's called move 37 through everybody. Question I have for health care. What's our move 37? Health care

is broken fundamentally. Do we have our own move that changes the way we think about healthcare? What's And we go to sovereign AI. Now we need to be more responsible for AI, healthcare, cyber security, pressures around data. Sovereign AI is a big thing now. In fact, UK has invested 500 million into sovereign AI over the last two months, building the infrastructure for a government and a country

to build their own platforms. Last slide is about the possible There's a machine that we create that works with us together. It contains we contain it. We control it. It allows us to reshape where we think about Ladies and gentlemen, it's really been a pleasure again speaking to you again after so many years. My book is out on Amazon. Uh feel free to uh get it and

read it and give me some feedback. Thank