Weighing the business potential of quantum AI

New trends in AI need to be carefully evaluated in terms of their benefits and challenges for businesses, and quantum AI is no exception. A burgeoning branch of the much-hyped field of quantum computerit uses quantum computing to implement machine learning algorithms.

Quantum computing is heralded as the future of calculation, but there is confusion about how it differs from traditional computing. For starters, while a bit is the most basic unit of information on a traditional computer, a qubit is the basic unit in quantum computers.

Because today’s quantum computers are small, they are impractical for performing most traditional computing functions. While traditional computing is likely to dominate most business applications in the near future, there are certain business issues that lend themselves to quantum approaches.

Quantum AI Business Strategies

Quantum algorithms provide a huge increase in computing speed to solve certain problems. Quantum AI, for example, can be very good at supporting machine learning, especially when that learning takes the form of neural network models.

While classical computing methods artificially create these hidden node models, they can be built naturally using qubits. The fundamental entanglement associated with multiplication, a mathematical tool for improving the accuracy of predictions made by a machine learning model, can be calculated much faster with qubits. This means that training neural networks on quantum computers can be orders of magnitude faster.

Another area is searching in unstructured databases, for which an increasing number of problems will arise as the internet operation of global calculations creates huge amounts of data. While classical computers are excellent at searching structured data, searching unstructured data is much less efficient. Lov Grover, an Indian-American computer scientist, developed a quantum algorithm that can guarantee a dramatic acceleration of searches. On small data sets, an acceleration is insignificant, but on large amounts of data, the practical accelerations are significant.

Consider the advantages and disadvantages of quantum AI

Benefits become apparent when quantum computers introduce algorithms that can solve problems exponentially faster. In practice, these “better” algorithms will only make sense if the problem space is large enough to justify the cost of quantum computers. As quantum machines get bigger and the price per qubit falls, the scope for solving these problems will increase.

There is no one-size-fits-all answer to whether using quantum AI makes sense for any given business. Business leaders must evaluate the costs against the benefits to determine when quantum AI will bring something compelling to their businesses. Three current drawbacks of quantum AI are the following:

Elusive ROI. Today, the ROI will be elusive in most cases. Even where the quantum algorithms prove to be faster, the costs are currently high enough to limit current investments to large governments, university research and large enterprises investments

Error correction is needed. Quantum data is very “noisy”. To get clean or “perfect” data, systems need to rest. Existing qubits do so at near absolute zero temperatures. The further you get from absolute zero, the more noise you will find in these systems. Therefore, for these systems to be viable, the systems must be built both at very low temperatures and with error correction as a fundamental part of the architecture. This means that such systems are usually even more expensive than experts expect.

Integration costs to classical systems. To be able to use quantum AI systems today, they must be integrated into the existing classical computing infrastructure, adding another layer of cost and complexity.

Quantum AI clearly holds huge promise for the future of business computing. However, current costs, complexities and niche algorithms remain major obstacles to the short-term adoption of quantum AI for most companies.

Leave a Comment

Your email address will not be published.