If you work in drug safety, you’ve probably asked yourself some version of this question: Will AI replace pharmacovigilance professionals?
It’s a fair question. AI in pharmacovigilance has moved beyond conference discussions and into everyday practice and professionals who know how to use it are already gaining an advantage.
In our series, The Future of PV: AI, People, and the Human Edge, Dr. Jan Petracek, our CEO, and Dr. Henry Seto, our Chief Medical Officer, tackle this question head-on. Their answer is both reassuring and thought-provoking.
Will AI replace pharmacovigilance professionals?
Probably not. The bigger risk is being replaced by someone who knows how to use AI better than you do.
Henry summed it up with a line he once heard on TV: “I’m not afraid of losing my job to AI, but I’m afraid of losing my job to another person who uses AI better.” Jan’s response was simple: Exactly.
The real threat isn’t the technology itself. It’s falling behind people who have learned to use it effectively. Jan also recognizes that many professionals are genuinely concerned about losing their jobs. He believes addressing these concerns is an important part of introducing AI into the workplace responsibly.
Why human expertise combined with AI outperforms either alone
According to Jan, the evidence points in the same direction: people working with AI can achieve better results than either people or AI working alone. The series begins with a similar observation: professionals who use AI perform better than those who don’t.
For pharmacovigilance (PV) teams, the takeaway is clear. The goal isn’t to hand the work over to a machine, nor is it to keep AI at arm’s length - it’s to bring human expertise and AI together.
What does it take to use AI effectively in pharmacovigilance?
Henry wasn’t always convinced of AI’s potential. For years, he was a skeptic, often taking the opposing position on conference panels discussing AI. His perspective began to change about a year to a year and a half ago when tools became available that were both effective and cost-efficient.
Since joining iVigee and gaining hands-on experience with these tools, he has seen their enormous potential but emphasizes one important condition: you must know how to use AI effectively.
If you don’t understand what you’re doing and simply expect AI to do the work for you, it won’t deliver the results you need. With the right knowledge and expertise, however, it can be an incredibly valuable tool.
Jan outlines several key elements organizations need to consider when adopting AI:
- An AI roadmap to establish a clear direction and set priorities.
- AI literacy training to help employees understand the technology and use it effectively.
- Implementation planning to integrate AI into existing processes.
- Fit-for-purpose tools that meet the organization's specific needs.
- Long-term scalability to ensure today's investments continue to deliver value as technology evolves.
Henry believes AI in pharmacovigilance is still far from reaching its full potential. In his view, we’re only at the beginning.
Why AI in drug safety still requires human oversight
Every powerful technology comes with risks, and AI is no exception. Pharmacovigilance professionals already specialize in risk management, so understanding and managing the risks associated with AI is a natural extension of their work.
At the heart of the issue are trust and transparency. Both are fundamental to drug safety, but AI cannot be relied on to provide either without appropriate human oversight. That’s why human supervision remains essential.
Jan sees firsthand how quickly the technology is evolving through his work teaching AI in pharmacovigilance, including a course for DIA that has been running for several years. The presentations, case studies, and exercises need to be updated every two months to keep pace with developments.
The demonstrations use systems such as Gemini and GPT, alongside specialized tools for literature searches, literature reviews, and medical reviews.
People are often surprised by how much these tools can accomplish. However, Jan stresses that realizing their potential requires an understanding of their limitations, including:
- Hallucinations: AI-generated information that sounds convincing but is inaccurate or fabricated.
- Model drift: Changes in a model's performance over time that can affect the reliability of its outputs.
- The need for verification: AI-generated information must be checked rather than accepted at face value.
To better understand how the technology works, the team even built its own retrieval-augmented generation (RAG) setup.
AI agents in case intake: real potential, real limitations
One of the newer developments discussed in the series is the use of AI agents, particularly in case intake systems.
Henry sees them as a significant step forward. Anyone who has worked in case processing knows how challenging it can be to enter large amounts of data into a database. He expects AI agents to make this process considerably more efficient.
However, he’s careful not to overstate their capabilities. Hallucinations and other limitations remain, and the technology still needs further development before it can reach its full potential.
The human advantage
The message throughout the series is consistent: AI in pharmacovigilance is neither something to fear nor something to trust blindly. It’s a technology to learn, use effectively, and supervise responsibly.
As the series puts it: Learn it. Use it. Lead with it.
The future of pharmacovigilance will need professionals who understand where AI adds value, how to use it effectively, and when human judgment must take the lead.
Watch The Future of PV video series above, then take an honest look at where your team stands today.
