After 2 years in stealth, a Cambridge, Massachusetts–based start-up has emerged with plans to conduct science in automated laboratories that run on a special version of artificial intelligence.
Lila Sciences has secured $200 million in seed funding from Flagship Pioneering, which incubated the company, and General Catalyst, March Capital, Ark Venture Fund, and others.
The company's overarching goal is to use “superintelligence” to solve humanity’s biggest challenges, such as finding new drugs and discovering greener solutions in the face of climate change, CEO Geoffrey von Maltzahn says. “It is, of course, a lofty mission,” he says.
The company’s algorithms are trained on both publicly available scientific papers and proprietary data generated by its automated labs. Lila aspires to create a centralized AI model that can aid in designing and performing thousands of experiments using robotics and automation across fields such as biology, chemistry, and materials science.
“Lila’s worldview is to increase the scale and sophistication of AI models that can unleash avalanches of new hypotheses that will be tested in a real-world lab to determine fact from fiction,” von Maltzahn says.
Although the company is still nascent, von Maltzahn says its platform is already helping solve real-world problems. For example, Lila scientists found a new catalyst for green hydrogen production with the potential to replace platinum-group metals like iridium, which are expensive and scarce. “The new catalyst compositions were discovered with Lila’s models in just 4 months, and it is a thousand times less expensive than the current standard,” he says.
Lila has already attracted some big names, including George Church, a Harvard University geneticist who is the company’s chief scientist.
Meanwhile, the company plans to use the funds it raised to scale its automated lab and advance superintelligence, which von Maltzahn describes as “one beautiful mind sitting at the center, which will benefit from organic, inorganic chemistry, and life sciences experiments.”