Intel 2020 Vision: The Evolution of Cognitive Computing

In the latest of our 2020 Vision series we examine how self-learning computer systems that mimic the human brain are helping businesses make more effective decisions

In the digital era, one of the biggest challenges for businesses across all sectors is the need to get more value out of enormous amounts of data. New technologies that allow us to sift through large amounts of unstructured data and draw useful conclusions, much faster than any human could, are increasingly important, and that's where cognitive computing comes in.

“The aim behind cognitive computing is not to replace humans entirely, but instead to augment human capabilities”

Cognitive computing refers to self-learning computer systems that replicate the cognitive processes of the human brain. It combines cognitive science with technologies such as Artificial Intelligence (AI), Natural Language Processing, visual recognition, neural networks, and deep learning in order to 'think' like a human.

Cognitive computing is expected to have a major impact on businesses in the near future, acting as a truly disruptive force in the new data-centric world. In fact, global spending on cognitive and AI systems is set to reach a massive $77.6 billion (around £59.9 billion) in 2022, according to analyst firm IDC.1

The aim behind cognitive computing is not to replace humans entirely, but instead to augment human capabilities. While AI alone is reliant on the people and the data that teach it, cognitive computing blends AI with other technologies to mimic the human thought process. And while AI is designed to take care of a very specific problem, cognitive computing studies patterns and makes recommendations based on its understanding of the data and context, without additional human input. The more data a cognitive system can access, the more accurate it becomes.

Cognitive computing allows businesses to differentiate themselves from the competition in several ways. It enables organisations to improve the quality of data insights, helping people to do their jobs more effectively. It also allows them to provide customers with better services at a faster pace, and also helps them to identify business opportunities.

Cognitive technologies are set to completely revolutionise entire industries in the coming months and are already making their mark in a number of areas. For example, AI-powered chatbots are helping to improve customer service interactions across various sectors while the finance industry is making use of AI-based systems to aid fraud detection.

AI is an integral part of cognitive computing and is required to make sense of the ever-increasing amount of data produced – and next-generation hardware is needed to support it. Intel recently unveiled its 2nd-Generation Intel® Xeon® Scalable processors, which are optimised for the most demanding processing requirements, including the flexibility to run complex AI workloads on the same hardware as existing workloads. The new chips also include a new built-in feature called Intel® Deep Learning Boost (Intel® DL Boost) to accelerate AI deep learning inference (with inference referring to its ability to apply existing knowledge to new data).

Intel DL Boost significantly accelerates inference performance for deep learning workloads such as image classification, speech recognition, language translation, and object detection. Taking embedded AI performance to the next level, DL Boost offers AI inference that is up to twice as fast as current-generation hardware.

At this year's CES in January, Intel unveiled the Intel® Nervana™ Neural Network Processor for Inference (NNP-I), a new version of its AI-powered chip. Developed with support from Facebook*, the new NNP-I chip is expected to go into production this year. While the NNP-I focuses on inference, a Neural Network Processor for Training, codenamed Spring Crest, is expected to be available later this year.

Also in development is Intel's neuromorphic test chip, codenamed 'Loihi'. The self-learning chip features circuits that mimic the human brain's basic operations with the aim of making machine learning more power efficient by combining training and inference in a single processor. The chip could eventually be used in any situation that requires the data to be processed in real time and could ultimately lay the foundations for cognitive computing platforms of the future.

Computers that can think like humans may still be some way off, but the foundations are already being laid. Cognitive computing can help businesses to make more effective decisions and with many organisations already integrating related technologies such as AI into their operations, those that don't embrace the future of cognitive systems risk being left behind.

*Other names and brands may be claimed as the property of others