Skip to main content

Tomorrow's medications could be invented by a deep learning supercomputer

PayPal won't let you buy prescription drugs
Developing new drugs to help cure diseases is a challenge some of the world’s sharpest minds are focused on. It’s also something that’s enormously expensive and time-consuming — with new drugs sometimes taking years, and costing hundreds of millions of dollars, to bring to market.

That’s where a United Kingdom-based company called BenevolentAI wants to help — and it’s using the world’s most advanced deep learning supercomputer to do so. The first company in Europe to use Nvidia’s state-of-the-art DGX-1 computer, the idea is to utilize cutting-edge chemical modeling algorithms to come up with ways to treat serious diseases faster than was previously thought possible.

“We’re taking giant corpuses of data, hundreds of millions of documents and structured data sources, and using it to discover relationships between chemicals, diseases and information about the body,” Derek Wise, vice president of engineering at BenevolentAI, told Digital Trends. “From that, we want to create a learning model that can help us predict more successful drugs — which work more effectively with fewer side effects — as well as [create] completely new, novel ideas for drugs that have never been attempted before.”

If anything, Wise explained, the drug discovery problem is getting harder than ever. Each day there are 10,000 updates on PubMed, the archive of biomedical and life sciences journal literature, which represents the sum total of new scientific knowledge being published. Being able to absorb all of this information is impossible for an expert in one discipline (or, well, anyone), which is why deep learning is so useful — since it can draw previously unseen correlations and create and test hypotheses astonishingly quickly.

“Yes, you can build artificial intelligence to play chess or checkers, but how is that really helping mankind?” Wise said. “The idea of BenevolentAI is really to help humanity. Pharmaceuticals are just one of those areas. There are lots of diseases and [therefore] lots opportunities to go out and help people.”

While the company will announce other applications for its deep learning system in the future, this is one that’s certainly going to be hard to top in terms of making a difference in people’s lives.

And, hey, isn’t there something pretty darn neat about the idea of popping a drug partially designed by an AI? Even if we’re 90 percent certain that’s how the machine takeover in The Matrix was carried out …

Editors' Recommendations