AI (artificial intelligence) is poised to become the next major general purpose technology that drives paradigm shifts in economic and industrial activity.
We already see its influence trickling into all aspects of everyday life, from smartphone assistants and robo-advisors, to multilingual customer service chatbots and investment priorities for national defence.
The AI technology is being adopted by developed countries.
But one wonders if less-developed countries can adopt AI and use it effectively?
“The answer is yes, but the scope is limited,” Huawei argues in its latest Global Connectivity Index (GCI) 2018.
“ICT enablers are layered, and each layer builds on the layers “below” it. Without sufficient broadband, you can’t deploy cloud. Without cloud, big data analytics is impractical. And without big data and IoT, you won’t have much clean data to feed your AI systems.”
However, the global firm further explains in the barometer that, you don’t need 100% 4G coverage and gigabit speeds to support cloud, and you don’t need data centres on every street corner to power analytics.
“There’s a degree of overlap, and understanding how each technology works, along with how it enables its peers, is key to knowing when and how to invest.”
As countries move up the ladder of ICT maturity, their ability to use AI – and the value they derive from it – will increase.
The following is a breakdown of each enabler and its relation to AI.
High-speed broadband is necessary for AI to function across an economy and provide value in all facets of everyday life.Broadband provides the connections that collect and transport data, distribute it for processing, and send instructions back to the smart devices – or people – who need it. Frontrunner countries have an advantage here, because they have the most expansive and high-speed broadband networks. Without sufficient broadband, Adopter and Starter countries run the risk of getting left behind. It goes without saying that, as the most fundamental component of ICT infrastructure, broadband is a necessary investment.
Most AI systems are installed, trained and run in data centres. Demand for data storage, micro-processors and servers have skyrocketed in response to AI requirements, reshaping the data centre industry.For instance, deep learning – a subset of AI – requires massive volumes of data to train, test and validate its neural network algorithms. GPU-based processing is ideal for these types of applications, so requirements for data centres that support GPU-based processing are growing.
Deep learning frameworks for AI create scalability challenges that can only be met by cloud services. Apart from governments, telecommunications companies and a few global IT providers, few companies are willing to spend the amount needed to scale out data centers to meet the growing demands of AI.
Building AI-capable systems at scale can be prohibitively expensive, largely because training algorithms require a huge amount of computing power. Cloud computing has vast stores of affordable computing power and storage that present cost-effective ways to build AI applications. Cloud with broadband connectivity is the critical infrastructure platform that enables AI adoption throughout an economy.
Big data is set to grow as we create and consume more digital content, and as daily life and business becomes more dependent on digital technology. We see explosive growth of data created by video for entertainment, industry, communications and security. The growth of IoT is massively contributing to data creation too. Every day, millions of gigabytes of data need to be stored, managed, tagged and processed for AI to make use of it. In time, this relationship will become a loop as analytics forms the investigative and learning capability for AI, and AI in turn improves the way that data is collected, managed and analyzed.
Internet of Things (IoT)
If AI systems are like a digital brain, then IoT devices are like the sensory organs that collect information, and the arms and legs that respond to decisions. IoT sensors provide AI systems with data to understand a given environment, and decisions made by AI systems are then carried out by IoT machines and other devices. An IoT sensor network and smart machines are necessary to unleash the full functionality of AI across an economy.