zapp2photo - stock.adobe.com
IBM is on a mission to enable developers, particularly those who build AI applications, to take advantage of the benefits of quantum computing.
IBM patented a series of quantum computing inventions in 2018, most of which involve hardware. But some of the patents help developers build apps, particularly those that use AI where the programs can take advantage of quantum computing systems, said Jeff Welser, vice president and lab director at IBM Research -- Almaden in San Jose, Calif.
The latest piece in IBM's quantum computing strategy is its updated IBM Q System One quantum computing system, designed for both scientific and commercial workloads. IBM also has built the cloud-based IBM Q Experience, which enables users to experiment with quantum computing, and the IBM Q Network, a community of organizations that work with IBM to advance quantum computing.
IBM also has updated its Qiskit open source quantum software development kit to help developers build, run and tune quantum computing applications, with specific topical areas of focus, such as chemistry and artificial intelligence, particularly machine learning apps, Welser said.
The power of quantum computing can meet the needs of complex AI applications that classical computing cannot, he said. Developers can take AI algorithms and map them on to quantum computing systems, then slice data along desired separation points to solve problems better and faster.
Mainstream quantum apps within two years? Probably not
Though quantum computing is in its infancy, IBM's Welser indicated that everyday use of quantum chemistry applications is about three years away and machine learning quantum applications might be as many as five years away. But it's not too soon to start developing algorithms and try to figure out the best to use and create apps for these systems, Welser said.
"IBM, Microsoft, Google, etc., need to give timelines for when quantum will start showing ROI, but in reality, we are not at a point where we can make such precise predictions," said Torsten Volk, an analyst at Enterprise Management Associates in Boulder, Colo.
Torsten Volkanalyst, Enterprise Management Associates
First and foremost, mainstream developers must have simple access to quantum computing resources to figure out how to build AI applications through quantum-specific capabilities, Volk said. The Qiskit Python library is a good first step, but there is still a long way to go until every developer can dabble in quantum to build AI applications, he said.
Quantum computing can especially help AI app developers accelerate machine learning training time, said Ronald Schmelzer, an analyst at Cognilytica in Ellicott City, Md. That's one of the most computationally intensive parts of AI, even more so than the actual implementation in real time of AI systems -- often it requires GBs of data to train multilayered deep learning networks, he said.
The quantum and AI patents are part of IBM's patent portfolio, which the company said includes 1,600 AI patents, 2,000 cloud computing patents and 1,400 cybersecurity patents in 2018. However, the number of patents does not necessarily reflect innovation.