Don't Fear The Machine – A Biologist Explains AI In The BioPharma Industry
By Louis Garguilo, Chief Editor, Outsourced Pharma
I should have known.
It takes an accomplished biologist to most effectively describe the significance of artificial intelligence in our outsourcing industry, and put it to practical use.
While building his career over several decades across pharmaceutical quality, and data-driven quality environments within the biopharmaceutical industry, Damiano Dragone was an early proponent of the promise of AI in development and manufacturing and the supply chain.
Today he continues that interest as Founder & Principal Advisor, Dragone Consulting, which he established after leaving Pfizer earlier this year.
For example, at Pfizer he was part of the manufacturing intelligence department, and worked with IT professionals developing AI systems for quality assurance.
“Back then, the technology specialists were developing AI-supported quality solutions,” he recalls. “The real challenge was never the technology itself. It was creating a common language between we quality professionals and those technology specialists.”
“We found the right way to speak to one another – simply stated, not leading with technicalities, and allowing for the mixing of our different kind of brains.”
This QA and IT/AI brain meld, and common language development, became essential, and while Dragone had opportunities to work on generative AI projects, he attended courses on AI.
He became a leader of interdisciplinary teams, and grew confident proposing what he felt would be interesting to have AI agents consider.
"Professionals should put together your industry and project experiences, successes and battle scars, and present that to the AI. Each of us remains the subject-matter expert of our domain," Dragone explains.
“By adding your core knowledge to AI agents you gain an awareness of possibilities to take into your quality, development and manufacturing, and supply chain practices.
“Generative AI can boost your creativity; it did mine. It does not replace expertise – it amplifies it."
The AI Iteration Plan
Just as Dragone learned to “speak the right way with the IT guys,” professionals must become adept at talking directly to “the machine,” as he often refers to AI models.
Consider, he suggests, professionals at a start-up biotech:
Chances are you cannot convene a roundtable of Nobel laureates or skilled colleagues from around the world – biologists and engineers to physicians – to solve a challenge related to the quality of your drug.
But if you “iterate effectively with the machine, it provides information and suggests avenues you nor those other humans may have considered.”
Important here is Dragone is not speaking exclusively about discovery or early development functions, but also manufacturing and supply chains.
This includes working with CDMOs, which we will get to subsequently. “The real value is not asking AI for answers; it's learning how to challenge its reasoning,” he says.
This becomes particularly valuable when sponsors and CDMOs need to make coordinated Quality decisions based on shared documentation and data, and scientific judgement.
Overall, says Dragone, when facing challenges, start an iteration process with AI systems. “I share my thoughts and experiences in detail, my way of approaching different problems.”
When AI provides its “non-human perspective,” you iterate further, ask more questions, and create a virtuous learning cycle to insert into the creation of something entirely new.
Approaching any element of work, he says, “my view is if you understand how to speak with the machine as the subject matter expert, you expand and improve your units of work and bolster creativity.”
“The machines work in a logic way, a stochastic approach, based on a logical sequence of non-human thoughts,” he says. Although he characterizes this as “talk,” Dragone points out that it is a mistake to anthropomorphize “the machine.”
“We produce from different perspectives and processes. This is positive. Otherwise, you might have to meet with those Nobel Prize winners around the table to do something so transitory.”
Don’t Fear The Machine
All this coming from a Quality-function-focused biologist may seem to raise some contradictions.
How can we trust and verify what AI agents give us? We’ve already heard of stories where overreliance on AI has caused serious troubles throughout society and industries, including with drug development and manufacturing outcomes at the FDA.
Dragone does not overlook nor is he unconcerned with any of the negatives applied to AI adoption (or overreliance). He insists we remain vigilant, and this too, we will discuss subsequently.
AI may support the reasoning process, but as Dragone stressed throughout our discussion, ownership of Quality decisions remain unmistakably human.
He is certain AI will greatly benefit our industry and patients who will receive more and improved drugs and therapies arrived at with its employ.
To reach fuller adoption, though, we must overcome “fear of change itself in our highly regulated industry.”
That fear, Dragone says, is overcome with a deeper understanding of the potential, for example, of how fast ideas and solutions can be arrived at and implemented, when compared to the past.
With best practices, including some outlined above, he says, “You will be impressed, and you will be excited. You may create that which you could not have in the past."
Stated another way, without utilizing these new AI tools, “maybe you do not discover the final cure for a certain cancer this decade.”
“With AI, I have experienced I can do most anything more quickly,” says Dragone. “The patient is then more quickly impacted by our life-improving and -saving drugs.
"Thinking of supply chains, I’m also convinced we can avoid drug shortages or chokeholds that our current systems fail to alleviate.”
Deviations, failed batches, investigations, analytical challenges – all should be addressed by both humans and the machine. “On the other side of the planet a patient is praying our drug will be there for them.”
Therefore, concludes Dragone, “considering ethical aspects, each company, each individual can add to our already virtuous way of following GMP standards and safety protocols when we learn how to effectively interact with AI tools.”
“This is something that can fill your heart,” he says.
“The challenge is for our industry to adopt AI and ensure AI-supported decisions remain scientifically sound, accountable, traceable and inspection-defensible.”
And that’s up to the humans, not the machines.