Hoped-for breakthroughs in superconductivity turned out to be duds, but there’s more in store, thanks to AI.

In the news

There’s no sense putting lipstick on this pig: A pair of developments around room-temperature superconductivity that we recently wrote about appear to be duds at best and frauds at worst. Room-temperature superconductivity has long been a scientific grail because of its potential to boost power grid efficiency, speed wired and wireless communication, make MRIs more affordable, enable real-world magnetic-levitation trains and more.

One of the developments included scientists from the little-known Quantum Energy Research Centre in South Korea claiming to have created a material that acted as a superconductor at room temperature. The alloy, dubbed LK-99, immediately became popular in the contemporary manner—that is, it went viral.

But when recognized supercomputer labs sought to test LK-99’s properties, nothing ensued but sad trombone. They found that “LK-99 isn’t a superconductor after all, and is actually a less efficient conductor than copper at room temperature.”

The other disappointment emanated from a University of Rochester team that made its own claim earlier this year that it had discovered a room-temperature superconductor. While exciting, this claim also generated skepticism, as team leader Ranga Dias had previously faced charges of research malfeasance. In recent weeks, Dias’s problems worsened when a respected journal announced it was retracting another of his papers (one not related to superconductivity) as “the result of an investigation that found apparent data fabrication.”

There is much to be said about today’s hype cycle and the pressure on academics to produce spectacular, social media-friendly results. But let’s focus on the end goal here: the pursuit of room-temperature superconductivity.

The Cognizant take

“While LK-99 might have been debunked, the search for suitable materials does go on,” notes Harvey Stotland, Strategy and Technology Industry Lead in Cognizant’s Communications, Media & Technology Segment. He points out that earlier efforts at room-temperature superconductivity focused on elements; broadening the search to alloys “obviously opens up a vast number of new possibilities. It’s the right direction of travel.”

Stotland adds that artificial intelligence, another technology very much in the news, may play a role in the eventual discovery of an actual superconductor. Today’s “typical path,” he says, features “lots of testing and experimentation with materials in similar classes to ones you know work to synthesize different materials and compounds.”

By contrast, “The AI path could ingest the electromagnetic, chemical and physical characteristics of known materials and use that data to predict the new compound’s properties, such as its pressure and critical temperature—that is, the temperature at which a material becomes superconducting. It could also be used to forecast the structure and characteristics of new candidate superconductors.”

Stotland warns that in this search, and indeed in virtually any large-scale use case involving AI (including generative AI), data curation and management must be front and center.

“The challenge is to build a corpus of data that could be used for both predictive and generative AI without introducing the kinds of errors that could be extrapolated into a candidate material that would be used for expensive physical experimentation,” he says. To that end, he adds, “finding peer-reviewed data and guaranteeing its sourcing should become a core competence of teams looking for these types of materials.”