Pundits have claimed the arrival of the “year of AI” many times over the last few decades. How well have those predictions worked out? But with today’s cheaper computing and more streaming data, artificial Intelligence (AI), created with analytics, may now be ready to take the field in mainstream enterprise technology. One exciting example, Dutch startup SciSports, is bringing (AI) and streaming data analytics to the world’s most popular game. The SciSports application relies on a new class of analytics that brings together the technologies powering the convergence of both AI and the Internet of Things (IoT). And in the process, it is turning volumes of raw data into powerful models and real-time information that make the “beautiful game” even more so for soccer clubs, their fans and coaches.
The SciSports software analyzes streaming video data from 14 cameras around the soccer pitch. It does live scoring of a deep-learning model to track the ball and all the players during a match. And it also relies on an in-memory platform to analyze huge volumes of data and train deep-learning models.
Applying AI and real-time analytics to big data on passing precision, player speed and jumping strength, for example, delivers a more complete game view. Fans can experience the game from any angle using virtual reality (VR). And coaches can better determine in-game strategy, such as timing of substitutions.
AI and IoT are complimentary
The SciSports application and many other examples demonstrate how AI and IoT are complimentary. IoT data sensors provide the raw material for AI-based applications and create enormous data volumes, often streaming. Today, much of that is unused or lost. Data without analytics is value not yet realized.
An analytics-based AI application sifts that data stream for insights and automates actions. And because AI is self-learning, it continues to find the insights that matter the most. The value is not in the data, it is in the insights the that AI reveals in it.
With new IoT applications creating a wealth of data, an AI approach might be the most efficient and fastest way to pull insights from the real-time data stream.
Why is this only now possible?
Until recently, the high cost of computing prohibited widespread use of AI and the high-performance analytics that power AI applications. The software was capable, but hardware and memory costs were prohibitive.
Today, powerful networks of computers – in house or in the cloud – are widely available and within reach of most IT organizations. And while AI itself is not new, what is new is its viability based on the current and future low cost of computing.
Now freed from hardware constraints, AI analytics add value by automating actions that previously were done by people. For example, if a financial firm wanted to optimize trades over the next 24 hours, it might take three regional experts 40 hours to refine the data, build the analytics and take action. An AI application can do the same thing in a minute.
Competing in the Analytics Economy
The world is increasingly driven by data and connectivity – and this change will grow dramatically over the coming years with the IoT increasingly churning out new data. According to some estimates, 45 percent of all streaming IoT data is never stored.
And the IoT will clearly generate even more data in the near future. Cisco predicts 50 billion connected devices by 2020. And Business Insider Intelligence forecasts that there will be nearly $15 trillion in aggregate IoT investment between 2017 and 2025. Yes, trillion with a “T.”
Technologies used in this new data-rich era, often referred to as the Analytics Economy, must be increasingly intelligent and automated. Enterprise technology organizations must learn how to build and manage highly intelligent analytics systems that use AI.
Connected devices are driving a new type of data set: an incredibly fast stream of valuable information that may never be stored. And the value of this data stream rapidly diminishes with time. Acting on the data at or near the point of origin and creation is optimal.
Volume and variety will move advanced analytic applications such as AI processing to the edge. IDC estimates that by 2019, 40 percent of all IoT-created data will be stored, processed, analyzed and acted upon close to or at the edge of the network; it will never reach the cloud or data center.
Data is a business asset because it can yield the insights needed to make better decisions. It is a matter of being able to understand the data, glean its insights and use them to answer questions. It is being able to discover things previously unknown. What if? Why?
IoT creates enormous streaming data volumes, much of it unused or lost. Data lost is opportunity missed. The key to unlocking the value of IoT data is with AI applications that self-learn and automate actions. AI applications – created with analytics and using real-time IoT data – can reveal new business opportunities long before the competition wakes up.
About the Author
Jason Mann, Vice President of Internet of Things (IoT) at SAS, is responsible for growing IoT revenue and providing global focus, strategic direction and alignment across the SAS IoT analytics portfolio. He oversees the research and development, product management and marketing, and execution of the sell-through strategy across the portfolio.
Sign up for the free insideAI News newsletter.
Speak Your Mind