Pharmaceutical leaders are assessing the integration and impact of artificial intelligence (AI) within the industry as regulatory frameworks evolve. Officials from both the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA) are focusing on risk-based validation and lifecycle management while emphasizing the necessity of human oversight in AI systems. However, specific AI frameworks remain in draft stages, requiring companies to adhere to existing good manufacturing practices (GMP) while preparing for future guidelines.
Concerns About AI Governance in Pharma
According to a recent study, only 17% of organizations have fully operational AI governance across their enterprise. This raises significant concerns regarding accountability, particularly as new use cases for AI continue to emerge. Key questions emerge about ownership of decisions, the authority to modify AI models, and the implications of incorrect AI outputs. Such issues are critical in a GMP setting, where gaps in governance could lead to integrity failures, uncontrolled changes, or performance degradations.
Experts indicate that traceability must be integral to AI systems from the onset, requiring detailed logging of actions, version control of models, and structured workflows for approvals. In an inspection scenario, it’s crucial for organizations to provide comprehensive evidence of the data used and the iterations of the AI model operational at the time.
Shifts in the Sponsor-CDMO Relationship
As pharmaceutical companies increase their collaboration with contract development and manufacturing organizations (CDMOs), the nature of these partnerships is evolving. Companies now focus not only on documentation exchanges but also on sharing manufacturing and quality data, which may be utilized by AI systems. It is essential for both sponsors and CDMOs to establish clear agreements about data ownership, access rights, and permissible uses of this data, particularly regarding training AI models.
Despite outsourcing manufacturing, sponsors maintain oversight responsibilities, and CDMOs must comply with GMP obligations. Quality agreements should delineate the responsibilities, documentation protocols, and communication methods regarding any necessary changes.
AI Utilization Differing by Development Stage
AI’s application in drug discovery has almost doubled, increasing from 17% to 35%, while its role in clinical and medical areas has remained stagnant. The disparity arises as discovery is characterized by high-volume, iterative testing, minimizing the costs of flawed ideas. In contrast, clinical and medical sectors face higher stakes and longer feedback cycles, typically requiring expert validation for AI-supported decisions. This discrepancy highlights the increased expectations for data quality and regulatory adherence in more critical development phases.
Notably, AI’s potential in analyzing process data, facilitating experimental planning, and enhancing scale-up processes is recognized, though overall adoption remains in its early stages. The interplay between development and manufacturing data is increasingly crucial to ensure that knowledge gained during development transitions smoothly into commercial production.
Changing Workforce Dynamics in Manufacturing
As AI technologies emerge, pharmaceutical companies are rethinking job roles, particularly in manufacturing and quality assurance. Nearly half of the value leaders surveyed are redesigning positions to incorporate AI support. For instance, quality assurance professionals might shift from labor-intensive tasks to strategic decision-making based on AI outputs. This transformation necessitates reskilling, emphasizing the need for precise training in leveraging AI systems effectively.
Moreover, advances in technology, such as augmented reality and virtual reality, are being leveraged to enhance training and operational efficiencies in manufacturing settings. Global centers, like Novartis’s site in Hyderabad, exemplify this trend, employing thousands in various roles including AI and operational support. As regulatory demands for oversight and accountability persist, the integration of AI will continue to shape the pharmaceutical landscape.
