Nvidia Says AI Can Finally Make Quantum Computing Work | PYMNTS.com

Before one can run a useful calculation, a team of specialists has to spend days manually tuning the system. Even after that, the machine keeps making mistakes faster than existing software can catch them. Banks running complex risk models and drug companies testing new molecules have been waiting on the same problem for years.

Nvidia thinks it has a fix. In a Tuesday (April 14) news release, the artificial intelligence company announced its launch of Ising, the world’s first family of open-source AI models built to solve those two problems: getting quantum systems ready to run, and keeping them running accurately. The models don’t change the underlying hardware. They make the hardware that already exists usable.

Why the Machines Keep Failing

Quantum processors pick up interference from their surrounding environment constantly, which throws off their calculations. Getting one ready to use has meant days of manual work — specialists reading output, finding where performance has slipped, and adjusting the system back into working order. Even once it’s running, mistakes pile up faster than existing software can fix them.

In a Tuesday blog post, Nvidia said the best quantum processors today make an error roughly once in every thousand operations. To become genuinely useful for business problems, that rate needs to reach one in a trillion.

Ising…