“Patients deserve more than to wait 12-15 years for a new drug to come to market. We need AI to change it,” according to Dr Patrick Schorderet, Global Head of AI Execution at Novo Nordisk Research and Development. In an interview for ICT&Health Global, he explains how the pharmaceutical industry is adopting AI and why the company plans to generate more data in 2026 than in the past 50 years.
The pharmaceutical sector is ranked surprisingly low in public trust, even below the oil and gas sector. Can artificial intelligence help regain confidence, and where should the pharmaceutical industry begin?
It’s interesting to see that the public – including the patients – see the pharmaceutical sector as being part of the problem, rather than part of the solution. It's humbling in many ways. It’s a cry for help. It’s a clear message that patients deserve more from us. The good news is we own the cards – and AI will help us be better.
For example, in many low- and middle-income countries (LMIC), the primary challenge is access to healthcare. People may live several hours from the nearest clinic or hospital, yet many own smartphones, which is a fantastic opportunity. Imagine a world in which, instead of traveling for hours, patients in remote areas would have access to AI-based diagnostics and care plans developed and vetted by the best physicians. A young girl in the Tibetan wilderness would have access to the same care as a wealthy individual living in London. Care is brought to the patient. It becomes closer. More inclusive. More trustful.
At the other extreme, while access to care is the main challenge in LMICs, high-income regions demand better access to innovation and greater transparency regarding costs.
Patients deserve more than to wait 12-15 years for a new drug to come to market. Pharma, together with the entire healthcare sector, needs to do better. The difficulty is that current processes for bringing drugs to patients have been overoptimized for more than a hundred years, so we shouldn’t expect change to come from there. However, over the past 10 years, with the advent of technology and access to high-performance computing, artificial intelligence has taken off, and has huge potential to drive the next paradigm.
On the transparency front, there is much to discuss, but let’s keep it for another interview.
Many pharma companies now have digital strategies. How mature is AI adoption in the industry? What works well, and where do you see limitations?
Pharma companies remain overall very entrenched in their own ways, which is the main reason that when technology disruptors appear, uptake and scale remain a challenge. Not because companies are unwilling, but rather because it is difficult to be something you are not. Everything about deploying cutting-edge technology innovation is new to the pharma sector. From the people you hire to the way you manage investments. Think about it this way: it’s not because you are an excellent runner that you are an excellent triathlete. You need fundamentally different skills.
On the bright side, being a good runner does help you because you have an understanding of training, nutrition, and commitment. The difference becomes how you apply that understanding to become an equally good swimmer and bike rider. It’s the same for pharma. We understand biology and chemistry like no other industry, but we are still just dipping our toes into AI. And it will take time.
The way we approach it is to have a clear strategy. Decide where you want to play, and how you want to win. The first question I always ask is: Is your objective to digitalize or to digitally transform? In simple terms, digitalizing is anything that makes you faster and more efficient. Microsoft Copilot is a great example. You don’t change the way you do things, but you get a lot better at the things you already do. Companies often start with this because it’s a lot easier. It’s closer to home. It's also not going to keep your company afloat for the next 50 years.
Blockbuster invested a lot in making the experience better. Automated search in their brick-and-mortar store. Online availability information. It didn’t save them from getting eaten by streaming services, which is how the movie rental industry digitally transformed. It’s about how you can cater to the needs of your customers in a fundamentally different way. At the end of the day, our patients want to get drugs that work. They don’t care how we get there.
This is an important conversation to have. Decide whether you want to do things better or do better things. And then commit to it.
You claim that cultural barriers are the main obstacle to scaling AI in pharma. What do you mean by “organizational culture in pharma, and how can it be addressed?
The best drugs are the drugs people take. Equally, the best digital solutions are the digital solutions people use. The challenge in technological innovation is that deploying it within an existing culture clashes with the culture itself. Let me explain: the culture of technology leaders, which de facto you need to bring in from the outside world, is fundamentally different from the company culture. They may see the world in the same way, but they speak different languages.
That is why it is essential to blend both worlds by developing solutions that solve business problems, not just cool technical artifacts. You need people who can bridge the gap. Energize both sides of the aisle. Be truly bipartisan leaders. Forcing co-ownership ensures the chiasm between technology and science closes, rather than deepens. I always tell my team to make sure that if something fails, someone in the business will care enough about it to call you up and brainstorm a plan B. The rest is chasing rainbows.
What is the “boomerang model of innovation” you presented at the “AI in Health” conference, and how does it explain the challenges of scaling AI?
The boomerang model is a simplistic representation of the path many companies take when trying to deploy technology in their business. It describes how digital maturity relates to organizational constructs. In summary, at the beginning, technological innovation happens decentrally. Think about the first person in your team who came rushing into your office to share their excitement for ChatGPT. The disruptors exist everywhere. It’s often messy because these people feel alone. Scattered. As the company matures, it seeks to capture synergies among these like-minded individuals by centralizing them. First into centers of excellences (CoE). Later, into innovation functions, sometimes even appointing a chief digital officer.
But as this group expands and grows – and hires externally – the disconnect with the business becomes more and more apparent. Timelines become slower. Suddenly, you need to open a ticket to get a simple thing done. The business grows increasingly frustrated until it decides to hire its own digitally savvy people to cater to their own needs. Innovation becomes once again decentralized while maturity grows, with the holy grail being a fully decentralized digital workforce. This is a generational question, though, so I’m not expecting this to happen overnight, especially in pharma.
The good news is that there is no good alternative. No shortcuts. So I would suggest being aware of the boomerang model, but don’t overintellectualize it.
Let us talk about data for AI. What practical steps should pharma companies take to prepare data for scalable AI?
When I speak to top leadership, I’m often asked what is needed for AI to make a difference, and my answer is always the same: Data. This triggers the same response: “We have more data than any competitor in the field, so we must win.” What companies often fail to understand is that, on the one hand, data has historically been generated for regulators to make a decision on whether drug A is better than drug B. This is how drug development works.
On the other hand, AI needs to be trained on a lot of data. It needs the reps. It needs the grind. And oftentimes, the amount of ‘historical’ data is not sufficient, and difficult to curate. Excel sheets or CSV files are the norm, stored on a personal computers somewhere in the world, with no global access. Outdated infrastructure and systems. How can it be that the most efficient way to share data in many global companies is to ask an associate to pack 20 hard drives in a suitcase and fly over to a site? This is not a parable. It’s unfortunately a real-life example.
In the age of AI, it’s essential that companies adapt their ways of working and start generating machine-ready data, not as an afterthought, but as a core principle and objective.
I’m also not saying that historical data is not useful – it can be. But in a world in which we can generate more data in one year than the combined 100-year legacy data pool, curating and preparing historical data, which is time-consuming and expensive, needs to be conscious, targeted, and serve a true business objective.
AI startups are entering the pharmaceutical sector with momentum. Is this the right path for innovation in the highly regulated industry with high safety standards?
The majority of drug companies will remain drug companies. Powered by AI. Not AI companies in themselves. Equally, the pace of innovation is far beyond what drug companies are used to. To put things into perspective, the time frame between chip generations or LLMs is roughly 12-18 months, while pharma’s is 5-10 years. This is why it is essential for pharma companies to build the muscle to be able to ingest and deploy external innovation. Rapid due diligence, pressure-testing, decision-making. Building this flywheel brings the best of both worlds. Focusing on what we do best while benefiting from the fast-paced technology ecosystem.
How do you imagine the pharma industry in 2040?
Many people have been saying we are at the tipping point for the past decade. I actually believe we finally are, for three main reasons. First, with massive investments into technology providers, wide access to high-performance computing is generally no longer an issue. Second, algorithmic architecture has made significant progress over the past 10 years, with advancements such as neural networks, transformers, language models, diffusion models, and multimodality, including chain-of-thought reasoning. Thirdly, automation and robotics allow us to scale data generation like never before, which is the cornerstone of AI’s impact.
While the technological revolution is here, the impact of AI will be driven by the people. This is the one thing that keeps me up at night. To see the benefits of AI for patients, we need to invest in developing multi-lingual talent – people who understand different languages, and who can bridge technology applications to real-world challenges. If we can scale this 100x in the next few years, which I’m confident we can, I genuinely expect that by 2040, the average pharma life cycle can be cut in half. Of course, as long as the rest of the industry adapts along the way, including regulators and payers.
