Automation now handles much of the heavy lifting, from adverse event reports to data structuring and analysis, but its limits are becoming clearer as adoption expands. Systems are effective at detection but less capable of understanding. For example, a rise in reports may reflect increased awareness rather than a new safety issue. Likewise, rare but severe events may remain overlooked due to low statistical weight. Statistics alone are not enough; contextual understanding remains essential.

This is why human oversight has become even more important. The role of the pharmacovigilance professional is shifting from processing data to interpreting it. Judgement, context and clinical reasoning are critical for safety decisions. AI can highlight patterns but cannot explain them with the nuance required for real-world application.

Bias is also an inherent issue. AI is only as impartial as the data it is trained on, and PV data has historically been uneven. Reporting varies by region, healthcare access and population groups, meaning algorithms may prioritise well-represented data while underweighting rare but serious events. Accepting AI outputs is easy; questioning them requires expertise and accountability.

As automation increases, human expertise becomes more strategically valuable. Professionals are expected to understand system limitations, validate outputs and ensure decisions remain transparent and defensible. The future of PV will not be defined solely by automation, but by how effectively people work alongside it. In a field where decisions directly impact patient safety, a gradual shift ensures the human layer remains both a safeguard and the foundation of decision-making.

“In Pharmacovigilance, ethics is ultimately expressed through human courage to question decisions made in the name of efficiency, automation or certainty.”

Vikalp Khare, Head of US GPV Safety Data Management at Otsuka