Department of Chemistry professor Anatole von Lilienfeld will be the editor of PRX Intelligence, a new American Physical Society (APS) journal focused on AI and machine learning methods and their application to advance scientific understanding.
The APS works to advance physics by fostering a vibrant, inclusive, and global community dedicated to science and society. It represents more than 50,000 physicists in academia, national laboratories, and industry in the United States and around the world.
The journal will accept submissions starting in February 2026, according to the launch announcement at the APS site.
Prof. von Lilienfeld is a Full Professor at the University of Toronto. The Ed Clark Chair of Advanced Materials and a CIFAR AI chair at the Vector Institute, Canada, he develops methods for a first principles-based understanding of chemical compound space using quantum mechanics, statistical mechanics, supercomputers, and machine learning. He is also interested in pseudopotentials, perturbation theory, symmetry, van der Waals forces, density functional theory, molecular dynamics, and nuclear quantum effects. He has authored or co-authored over 130 peer reviewed publications and has a Google Scholar h-index of 68.
His research has received external support including grants from NSERC, CFREF, and CFI in Canada, an ERC consolidator grant in the EU, and an assistant professorship grant by the Swiss National Science Foundation.
His career as an editor has been distinguished. In addition to this new role as Chief Editor of PRX Intelligence, von Lilienfeld has served as the inaugural Editor in Chief of MLST (IOP Publishing), as an Associate Editor for Science Advances (AAAS), JCTC (ACS) and JACS (ACS), on the Editorial Advisory board of Scientific Data (Nature), and as a Senior Scientific Advisor for the machine learning journal series published by IOPP.
“The speed and scale of innovation happening in the scientific machine learning and AI space has been explosive. We’re excited to support these innovations by providing a new high-impact, open access platform for work that is fundamental to the future of science and society,” said Rachel Burley, APS’ chief publications officer, in the ACS announcement of the journal’s launch.
Chemistry Stories is excited to see the story of this new journal unfolding in realtime. We caught up with Prof. von Lilienfeld to ask him for the inside story on PRX Intelligence, the dynamic AI research landscape, and the mission of making science as widely accessible as possible. Here’s what he told us:
What inspired the launch of PRX Intelligence, and how did you come to be involved?
Science started with experiments. To explain the observations, theory developed with the goal of providing a quantitative understanding, and we arrived at theoretical frameworks that contained equations. However, some were too difficult to solve analytically.
When the digital revolution set in, a new paradigm emerged: computer simulations enabled numerical solutions of some of these difficult equations, typically in approximate forms.
Thanks to Moore’s law, we have enjoyed increasingly easier access to computation over the last decades, and many of the simulations as well as sophisticated experimental results were amassed for more instances than ever.
Combining and complementing these three pillars of science (experiment, theory, and computation), sufficiently large data sets were obtained, making more aspects of the entire field amenable to the statistical learning paradigm. As a result, over recent years we have witnessed the emergence of AI as a fourth pillar of science - in addition to experimentation, theory, and computation.
It is in this context that I view last year’s Nobel prizes in physics and in chemistry as appropriate, and the decision to launch a new journal dedicated to AI for science to be most timely.
How do you envision its role in shaping the future of AI and machine learning in scientific research?
PRX Intelligence is an interdisciplinary journal which welcomes all experimental, theoretical, and computational physics research from all the branches of the physical sciences and whenever it has benefited from statistical learning, or, conversely, whenever physics-based principles have led to novel AI insights or methods.
AI in science has become so common right now that unfortunately in many papers it is not always clear how general and significant the findings are. Within PRX Intelligence, it is our goal to uphold (or even raise) the standards and quality of the APS family of journals while trying to attract and to identify key studies that truly represent major new insights or advancement, as well as to relate those advances to the physics space in the best possible way.
In particular, we will foster an inclusive culture within a journal run by scientists for scientists, where hand-picked editors and referees do full justice to the authors’ work. All our editors and editorial board-members are active Principal Investigators, and we will try hard to match the submitted manuscripts with the most competent referees within a constructive process that results in the best possible content.
AI and machine learning are transforming many fields. How do you see PRX Intelligence being a part of this transformation?
Within the theoretical and computational physical chemical and materials sciences, a large fraction of the community has focused over the last decades on establishing a solid and robust understanding that allows more and more of our computational approaches and approximations to be predictive and feasible. The hope is that, given enough data and sufficiently advanced machine learning architectures for training, we will be able to improve the situation to a degree that experimental planning and autonomous experimentation and characterization become routine.
This would be a tremendous benefit for the accelerated design and discovery of new molecules and materials, e.g. catalysts or semiconductors.
Of course, these are typical problems just for the field I am coming from. However, I hold the strong belief that this trend is even more general and that most, if not all, branches of the physical sciences stand to benefit in similar ways, because statistical methods successfully manifest themselves in a variety of approaches throughout the physical sciences where falsifiability, control experiments, causal relationships, stochastic regimes, and emergent patterns prevail.
Consequently, I believe that it is clear that the entire plethora of modern statistical learning can be applied, including unsupervised and supervised learning, generative and reinforcement learning, foundational models, LLMs, agentic AI — a lot of new tools that have and will continue to become ever more useful.
Why was open access chosen for PRX Intelligence?
This was a decision made by the publisher. But it is in line with my personal philosophy which has always been to make heavy use of arXiv and open source, and to not hide research work behind paywalls. I believe that this is particularly important for scientists in countries where academic libraries cannot necessarily afford the subscription fees. We want people from around the globe to be able to access the content published, and we believe strongly that sharing scientific insights openly is a deeply humanistic act that helps humanity as a whole advance. This is something that the open access model enables.
What advice would you give to researchers who want to contribute work in February?
I certainly view my personal research interests in the chemical and materials sciences as representative of the kinds of readers and authors we’d like to address within this journal. PRX Intelligence is an interdisciplinary journal where we consider work from researchers in all domains of physics related science.
The papers should somehow combine AI and machine learning with theory, simulation, and/or experiments. We would like to especially invite those who combine data with physics and statistical insights to develop models which lead to new insights.
As you look forward to launching the journal, what impact would you like to have?
I look forward to working with the editorial team to champion rigorous, impactful research at the intersection of machine learning, AI, and the physical sciences,. By connecting researchers, practitioners, and engineers interested in AI and machine learning, our goal is to support the communities that could foster cross-disciplinary collaborations, potentially leading to cutting-edge scientific discoveries.
We believe strongly that sharing scientific insights openly is a deeply humanistic act that helps humanity as a whole advance. -Anatole von Lilienfeld