Parilov/Adobe Stock

Artificial intelligence is quickly becoming part of daily life, and AI’s tangible advances during the past year earned 2023 the title of the “Year of AI Transformation.” The most popular AI applications include OpenAI’s text-based ChatGPT and image-based DALL-E 2, tools that have caused notable controversies in arenas like education. However, despite concerns ranging from cheating to reporting misleading data, the potential benefits of AI are significant—particularly as AI expands into new territories.

One growing sector for AI adoption is materials science. Scientists have begun exploring the use of AI tools to automate a variety of tasks that are typically time-consuming, including manual endeavors such as identifying relevant references or potential compound ingredients. Remarkably, researchers have also begun employing AI to discover new materials and develop current material compounds in previously unexplored ways. These demonstrated advantages suggest that AI will play an increasingly fundamental role in the creation of new materials as well as the optimization of existing ones.

For example, scientists are increasingly employing tools like ChatGPT to function as research assistants. A typical arduous task for researchers is the review of existing scientific literature. A recent article in the Journal of the American Chemical Society documents the research team’s experience using ChatGPT to streamline this effort by mining the text of more than 200 papers and 26,000 specific material factors. The team set out to determine potential clean energy applications for metal-organic frameworks, and ChatGPT found 800 different MOF compounds referenced in the peer-reviewed articles. Furthermore, the group created a new machine-learning model from this data as well as a chatbot interface—which it calls the ChatGPT Chemistry Assistant—allowing narrative, language-based data retrieval for those without coding experience.

In England, scientists at the University of Liverpool’s Department of Chemistry and Materials Innovation Factory developed a new AI tool to discover new materials. Given the many elements in the periodic table, there is a near-infinite number of potential compounds, including previously unknown iterations. According to lead chemist Matt Rosseinsky, material scientists often set out to find new compounds that resemble existing substances, but this approach rarely leads to novel discoveries. The new AI app “combines the ability of computers to look at the relationships between several hundred thousand known materials, a scale unattainable for humans, and the expert knowledge and critical thinking of human researchers that leads to creative advances,” Rosseinsky explains. The team’s new tool has thus far enabled it to discover four new solid-state materials that it anticipates will lead to improved battery technologies.

Researchers are also employing AI to design materials, emphasizing specific priorities for enhancement in known applications. For instance, a team at research organization Max-Planck-Institut für Eisenforschung in Germany has used a language-processing tool similar to ChatGPT and machine learning approaches to facilitate the development of more corrosion-resistant alloys. This research is obviously well-suited to architecture and construction. The global financial impact of corrosion is significant, reportedly totaling more than $2.5 trillion annually, thus motivating improved corrosion resistance in alloys and coatings. “We trained the deep-learning model with intrinsic data that contain information about corrosion properties and composition,” physicist and team leader Michael Rohwerder explained in a university press release. “Now the model is capable of identifying alloy compositions that are critical for corrosion-resistance even if the individual elements were not fed initially into the model.” According to the researchers, the new method demonstrates a 15% improvement in predictive accuracy compared with previous approaches.

These are a few examples of the rapidly growing adoption of AI in materials science. Other endeavors include the development of oxidation-resistant coatings, shape-memory alloys, magnetic materials, and optimized electronic structure models. As these AI-enhanced experiments in materials research and design become more developed and codified, they will open up possibilities for adoption in additional areas of materials science where significant needs exist—such as renewable power, energy storage, recyclable compounds, or phase-change materials.

In fact, we can query ChatGPT about such priorities. To my prompt: “What are the most significant areas of need for advances in materials science based on ecological and resilience imperatives?” ChatGPT corroborated some of my suggestions above and added: “Developing biodegradable materials, improving resource efficiency, and creating materials that withstand extreme conditions are also crucial areas of need.”

The views and conclusions from this author are not necessarily those of ARCHITECT magazine or of The American Institute of Architects.

Read more: The latest from columnist Blaine Brownell, FAIA, includes looks at immersive art exhibitions, material technologies that improve air quality, and reducing the carbon footprint of bricks.