Overview
The overall goal of this project, co-funded by a financial gift from Google, is to create a delocalised global ‘Hive Mind’ that directs autonomous laboratory robots to discover engineered porous materials for atmospheric CO2 capture. We will fuse human insight and AI agents with experimental and computational data streams in real-time, closed-loop robotic experiments to build a new paradigm for tackling complex societal challenges. This studentship will focus on the development of “chemically-aware” agentic AI methodology that can orchestrate autonomous discovery, acting as the ‘brain’ for the robot chemist.
About this opportunity
The threat of rising carbon dioxide (CO2) levels is well documented. Our goal is to create a delocalised global ‘Hive Mind’ that directs autonomous laboratory robots to discover engineered porous materials for atmospheric CO2 capture (a.k.a. DAC) using multiple data modalities. We will fuse human insight and AI agents with experimental and computational data streams in real-time, closed-loop robotic experiments to build a new paradigm for tackling complex societal challenges beyond DAC. In a strategic project co-funded by Google, we are proposing to fuse real-time experimental data from AI-powered laboratory robots with global, crowdsourced human expertise, AI agents, and computational predictions to create a unified, hybrid intelligence—a Hive Mind—powered by these four data modalities. We believe this will be necessary because of the exceptionally difficult scale-up challenges for DAC materials, which will defeat ‘brute force’ robotic strategies, human knowledge, AI agents, or large-scale computation used in isolation. It will also demonstrate a new approach to global cooperative research that is relevant to problems beyond DAC.
You will work in a unique interdisciplinary research environment since our team covers all the sub-areas required to tackle this challenge, including porous materials for CO2 capture (J. Am. Chem. Soc., 2025, 147, 23160), autonomous mobile robotic chemists (Nature, 2020, 583, 237; Nature, 2024, 635, 890), machine reasoning using LLMs (IJCAI, 2025, 4967), and ‘human-in-the-loop’ approaches (IJCAI, 2024, 3881).
The student will develop an agentic AI framework – essentially the ‘brain’ for this Hive Mind approach, which will autonomously and adaptively accommodate feedback from multiple data modalities towards a more accelerated search for porous materials. In addition to the overall framework the student will further develop novel DAC-specific AI discovery tools to be used ad hoc by the agentic framework to facilitate faster and better discoveries.
This project will be supervised by Prof Andrew Cooper FRS (Department of Chemistry), Prof. Katie Atkinson (Department of Computer Science and Informatics), Dr Xenofon Evangelopoulos (Department of Chemistry), and Dr Gabriella Pizzuto (Department of Computer Science and Informatics & Department of Chemistry). The supervisory team combines experts in artificial intelligence (Prof. Atkinson and Dr Evangelopoulos), automation for materials discovery (Prof Cooper), and robotics (Dr Pizzuto).
Essentially all the hardware tools required for this project have already been built within the group (synthesis robots, fast screening for CO2 sorption, etc.). The day-one challenge will be the development of a ‘discovery AI tool’ that intelligently integrates them with machine reasoning (e.g., from LLMs) and a panel of human experts’ feedback. This studentship will focus specifically on the AI framework for coordinating and directing the experiments.
This project is expected to start in October 2026 and is offered under the EPSRC Centre for Doctoral Training in Digital and Automated Materials Chemistry based in the Materials Innovation Factory at the University of Liverpool, the largest industry-academia colocation in UK physical science. The successful candidate will benefit from training in robotic, digital, chemical and physical thinking, which they will apply in their domain-specific research in materials design, discovery and processing. PhD training has been developed with 35 industrial partners and is designed to generate flexible, employable, enterprising researchers who can communicate across domains.
Further reading
Cissé, et al. HypBO: Accelerating black-box scientific experiments using experts’ hypotheses. Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence (IJCAI-24), 3881–3889 (2024). https://www.ijcai.org/proceedings/2024/0429.pdf
Cissé, et al. Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence (IJCAI-25), 4967–4975 (2025). https://www.ijcai.org/proceedings/2025/0553.pdf
Burger, et al. A mobile robotic chemist. Nature 583, 237 (2020). https://doi.org/10.1038/s41586-020-2442-2