Artificial intelligence is hunting for the science we've overlooked

Filtering and categorising all of human research

About two million scientific papers are published every year. In that deluge of research, it's almost certain that we're missing things, so an artificial intelligence has been tasked with crawling through it to spot the good bits.

'Semantic Scholar', as its creators have named it, has the ability to read, digest and categorise publicly-available research papers. It scans text and images, identifies references and citations, and can even determine which papers are more influential or controversial.

What's Going On

It's the work of the Seattle-based Allen Institute for Artificial Intelligence (AI2), and will begin work on Computer Science papers before branching out to biology, physics and other hard sciences. It joins other, similar projects like IBM's Knowledge Integration Toolkit and DARPA's Big Mechanism.

"Our vision is of a scientist's apprentice, giving researchers a very powerful way to analyse what's going on in their field," Oren Etzioni, director of AI2, told New Scientist.

"Ultimately, perhaps a human scientist doesn't have to read it at all."

Researchers have warned of the dangers of an augmented reality future