Artificial intelligence-enabled self-driving laboratories in bioprocess engineering are most likely to develop as hybrid systems that integrate automated decision-making with sustained human oversight, according to a recent review.
The authors examined applications of artificial intelligence (AI) and laboratory automation in bioprocess development, outlining technical, regulatory, and scale-related considerations that distinguish bioprocessing from chemistry and materials science.
The authors described levels of laboratory autonomy ranging from Level 0 (no autonomy) to Level 5 (AI scientist). They reported that most current bioprocess automation operates at Levels 1 to 2, reflecting substantial human oversight due to biological complexity, experimental variability, safety requirements, and regulatory constraints.
The review proposed a modular hybrid-lab framework consisting of fully automated core processes (autonomy Levels 3 to 5) combined with auxiliary processes that may be partially automated or manual. Core processes can include automated strain construction, culturing, sampling, and absorbance-based product measurement, with experimental data stored in databases for AI–guided iteration. Auxiliary modules such as transcriptomics, proteomics, metabolomics, or alternative strain-building strategies can be integrated as needed without altering the core workflow.
Recent developments cited in the review include large language model-based tools for experimental design and protocol translation, computer vision for laboratory monitoring, and multifidelity optimization approaches that integrate robotic and human-generated data. The authors stated that hybrid systems should incorporate defined decision tiers, required human checkpoints when measurements fall outside expected ranges, and auditable activity logs aligned with regulatory requirements.
Scale-up was identified as a primary challenge. While self-driving laboratory concepts have been applied at experimental volumes of approximately 1 to 250 mL, translation to pilot and manufacturing scales of approximately 50 to 200,000 L involves nonlinear process behavior, changing bioreactor dynamics, and differences in monitoring and control infrastructure. The review discussed digital twins and Bayesian optimization frameworks, including multifidelity modeling, as approaches to support scale transfer.
The authors also identified data standardization, shared protocol languages, and cross-institutional benchmarks as necessary components for broader implementation of self-driving laboratories in bioprocess engineering.
The authors wrote that “the next frontier is not the replacement of scientists but the careful design of hybrid systems, where human expertise and AI-driven automation complement one another to enable reliable, scalable, and responsible bioprocess innovation.”
Newsletters
Receive the latest pharmaceutical news, personalities, education, and career development – weekly to your inbox.
