The Enormous Potential and Challenges of AI and Machine Learning

 

McKinsey Global Institute just released a discussion paper analyzing the practical applications and economic potential of artificial intelligence (AI). Notes from the AI Frontier looks at hundreds of AI use cases across 19 industries to determine where AI, machine learning and neural networks could create the greatest value.

The numbers are mind-boggling. McKinsey estimates that AI has the potential to add $3.5 trillion and $5.8 trillion in annual value to the global economy. This represents the revenue growth, cost savings and enhanced products and services that are enabled by AI technologies. Some organizations and even entire industry sectors are likely to see a radical shift in market dynamics as AI adoption becomes more widespread.

The report notes that AI can also improve existing data analytics applications in areas such as predictive maintenance, logistics optimization, customer service management and personalized marketing. In fact, 69 percent of the opportunities to improve performance with AI involved existing analytics use cases, with just 16 percent representing new applications involving AI alone.

Despite its potential, however, AI comes with significant challenges. It requires the ability to analyze massive data sets — much larger than those used in more traditional analytics applications.

Performance for AI Workloads

AI workloads require high performance and considerable capacity to rapidly manage and analyze huge volumes of structured and unstructured data. To address this challenge, Dell EMC has just introduced two new four-socket servers that help organizations leverage AI and machine learning to gain intelligent insights and achieve better business outcomes. Part of the 14th-generation Dell EMC PowerEdge server portfolio, the servers combine the exceptional performance density of Intel Xeon Scalable processors (up to 112 processing cores) with up to 6TB of memory and NVDIMM options.

The Dell EMC PowerEdge R940xa is designed to accelerate databases for business-critical applications without cloud fees and security risks. It improves application performance by combining up to four CPUs with four graphics processing units (GPUs), and enables low latency with direct-attached non-volatile memory express (NVMe) drives.

The Dell EMC PowerEdge R840 is designed for in-database analytics. It minimizes latency with more direct-attached NVMe drives than other options available on the market today, and speeds data transfers with a fully integrated ultra-path interconnect (UPI).

Integrating Intel FPGAs

Dell EMC has also integrated Intel Xeon Scalable processors and Intel Arria 10 GX FPGAs into its PowerEdge R740 server to create the Dell EMC Ready Solution for HPC (high-performance computing). Dell EMC Ready Solutions are tested and validated platforms that help accelerate innovation, reduce risk and lower total cost of ownership.

Intel FPGAs (field-programmable gate arrays) can be customized to accelerate specific workloads, and enable developers to turn off unused features to minimize power requirements. FPGAs are increasingly used for AI, machine learning and other applications that benefit from massively parallel processing. In addition to the PowerEdge R740, the Arria 10 GX FPGAs are available in the PowerEdge R640 and R740XD servers.

As a Dell EMC Titanium Black Partner, FusionStorm has the knowledge and certifications to help you take advantage of these powerful new solutions. If your organization is looking to capitalize on the value potential of AI, we can help design and implement the right server platforms to support your AI initiatives.