At first glance, there’s nothing out of the ordinary about the Acceleration Consortium laboratories at the University of Toronto. It seems like business as usual, in line with the other science labs on campus: test tubes, robotic arms behind plate-glass barriers, machines that light up and hum, and grad students hunched over projects. It’s what you can’t see that makes this unassuming space extraordinary. Every last test tube, robotic arm and machine is self-driving, and artificial intelligence is behind the wheel. The result: facilities so advanced, they’re redefining how scientific research is conducted.
Alán Aspuru-Guzik, a professor of chemistry and computer science at U of T, heads up the Acceleration Consortium, a group of academics, government institutions and industry leaders around the world. He and his colleagues are working to identify molecules and materials that could be crucial to solving the world’s biggest challenges. Aspuru-Guzik set up these self-driving labs to identify, develop and test promising new materials faster than humans ever could. The process of discovering novel materials is painstaking work, requiring years of careful research. He doesn’t want to wait that long.
“On average, a material goes from idea to market in 20 to 25 years,” says Aspuru-Guzik. “These AI-powered self-driving labs speed up that process and bring down the cost.” Automating certain tasks can reduce development time to a matter of a few years and shave down the average $100 million price tag to $1 million, he says. For instance, earlier this year one of the consortium’s self-driving labs developed a potential treatment for liver cancer in less than a month.
Think of this technology as akin to a self-driving car. Instead of punching in an address on the dashboard screen, scientists feed an AI program details of what they’re hoping to produce, whether it be a new medicine, biodegradable material or construction industry coating. The AI program is hooked up to the robotic lab equipment, which then synthesizes these compounds, analyzes results and runs the experiments — at speeds no human could match. This process repeats 24 hours a day, seven days a week, as many times as needed, until a new material is successfully developed.
Self-driving labs in the Acceleration Consortium are developing new drug formulations for more effective cancer treatment, materials that can convert waste CO2 to valuable products, and the next generation of safe and low-cost batteries. In just two months, Aspuru-Guzik and his team tested more than a thousand molecules for potential use in organic lasers, or OLEDs. Before the advent of AI and self-driving labs, he says, “it took researchers in the field five years to publish 10 laser molecule candidates.”
But Aspuru-Guzik has no interest in starting a pharmaceutical company or manufacturing solar panels; the Acceleration Consortium is squarely focused on uncovering potential. “Once the material is there, innovators will come up with what to do with it.”
His colleague Jason Hattrick-Simpers, an engineering professor and one of the newest members of the Acceleration Consortium, is using AI to develop and test new wear-resistant alloys that could fight off rust and corrosion.
“Cars, planes, buildings, powerlines, concrete, pretty much any material is constantly battling the environment and oxidizing and degrading,” Hattrick-Simpers says, citing the roughly $70 billion that Canada spends every year to mitigate the effects of corrosion. While it’s impossible to fully prevent these materials from wearing down over time, the process can be mitigated. “This could be a game changer for improving the longevity of offshore wind turbines, which are exposed to all that moisture and salt, or the mining tools used in Canada’s north,” he notes. “They need to resist the elements to be useful.”
The advanced lab tool that makes this happen is something called the Sputtertron. This device blasts the surface of a material with energetic particles to coat one material with another, a technique known as “sputtering.” Its AI system analyzes the new alloys produced and figures out how to make them even stronger during the next round of tests. The device is able to churn out about 24 new physical samples a day. Compare that to the two or three samples Hattrick-Simpers’s team could produce before they had the benefit of this technology. “And those were long days,” he adds.
Similar work is being carried out in self-driving labs at the National Research Council of Canada’s advanced materials research facility in Mississauga. A partner of the Acceleration Consortium, the facility focuses on helping Canada meet its emissions reduction targets. Researchers there are working on a swath of projects, including clean-energy materials and batteries as well as complex oxides and alloys, which could advance the next generation of energy storage materials.
With $519 million in funding, $200 million of which came from the Canada First Research Excellence Fund, the Acceleration Consortium is looking to hire and train new scientists and researchers, further develop AI lab technology and update existing facilities. In addition, six new state-of-the-art self-driving labs will be built on campus and another at the University of British Columbia over the next two years. Each lab will have its own specific focus, from carbon capture and flow batteries to nanomedicines and implantable sensors.
Aspuru-Guzik is already thinking about what’s next. He has two big priorities for self-driving labs: standardization and democratization. Harnessing the full potential of this technology will happen, he says, when it expands beyond futuristic labs to which only a few hold the key.
“My goal over the next five years is to make self-driving labs ubiquitous,” he says. “Just like 3D printers are available at libraries, I want to make this technology available to everyone in the world.”
Photo credit: Acceleration Consortium, University of Toronto