Jyunmi Hatcher

King's own robot, Eve, is a platform at Chalmers University that takes up half the floor space of its laboratory.

Five meters square, three meters high.

Eve nominate automates early stage drug design.

One of its notable results came when it independently screened approximately sixteen hundred chemicals and modeled how their structures related to their biological activity.

and identified that triclosan, a common antimicrobial compound, could target an enzyme critical to the survival of malaria parasites during their dormant liver phase.

The robot didn't just run the assay.

It designed the experiment, performed it, and arrived at the finding that gave researchers a potential new route to fighting treatment-resistant malaria.

King is now building a successor called Genesis, which he estimates will cost roughly one million pounds to construct the same price as Eve.

But it'll be at least 10 times cheaper to operate than the human labor for equivalent experimental output.

Genesis will take about 10,000 mass spectrometry measurements per day.

At the University of Toronto, a chemist, Alain Aspuru-Guzek, leads the Acceleration Consortium, a fleet of more than 50 self-driving robots spread across multiple labs and universities.

It's backed by the largest federal research grant ever awarded to a Canadian university.

$200 million in Canadian dollars, with matching commitments from partners bridging the total investment of roughly half a billion dollars.

The consortium's goal is to reduce the time and cost of bringing new materials to market from 20 years and $100 million to as little as one year and $1 million.

Their robots are working on everything from cancer drugs to low-carbon cement to biodegradable plastics.

One of Aspiro Guzik's former postdocs, Gabe Gomes, at Carnegie Mellon University, built a system called CoScientist that represents the next generation of these platforms.

Unlike earlier systems that rely on specialized optimized algorithms, CoScientist is driven by a large language model, in particular GPT-4, that can interpret scientific problems described in plain English.

It'll also search the web and look at scientific literature for relevant information.

It'll also plan experiments and interface with robotic hardware to execute them.

Gomes describes the system as field agnostic as the AI models improve the range of problems it can tackle expands with them.