This week at the Intel Shift Conference in New York, I had the opportunity to listen to my colleague Amir Khosrowshahi, CTO of the Intel AI Products group, speak to a gathering of business executives about the transformative impacts of AI.
Amir explained how artificial intelligence (AI) can change what organizations do and how they do it, creating new business opportunities. Every company is in some phase of their AI adoption course: evaluating and understanding the opportunities, testing AI use cases and its outcome on their business, or fully integrating AI systems that are increasingly driving business metrics.
AI concepts have been around for more than 60 years, but we now have the technology to make AI a reality. AI is predicated on the simple idea that with the right training a computer can simulate human decision making. It processes huge quantities of data to build a model that can draw inferences from the data and use those inferences to predict outcomes, take actions, or make “decisions.” Machine learning requires two things: lots of data and lots of computational power. We have both of those now.
Thanks to the Internet and to the instrumentation of everyday devices, data is exploding. By 2020, it’s projected that a single Internet user will generate about 1.5 gigabytes (GB) of data every day, a smart car will produce 4,000 GB of data daily, and a connected factory will generate 1 million GB. And I’ll bet if you look around in your business you’ll find that data is available to you in volumes that were unthinkable just a few years ago.
The second requirement for AI is vast computational power. The computing capability needed really hasn’t been available until the past few years. But now Intel is developing specialized processors designed for AI workloads, our Intel® Nervana™ Neural Network Processors, and we are building machine learning capability right into the Intel® Xeon® processors that already power the majority of data centers around the world. This processor power is really only needed for the learning process (“training” in AI terms). Training produces a much smaller “model,” which can be easily deployed into factory floors, hospitals, mobile devices, or even self-driving vehicles to let them apply what has been learned to take actions based on similar data.
As we learn how to apply AI, applications are proliferating. AI techniques power image recognition apps that are increasing security and improving transportation systems in smart cities. AI is helping conservationists identify and track individual whales and polar bears, and it’s helping to identify and find missing and exploited children faster. AI is already building automated response into vehicles and will enable completely autonomous cars, trucks, and aircraft in a few years. Online retailers are using it to suggest products that match your buying history, preferences, and current needs. And manufacturers—including Intel—are using it to increase yield and improve product quality.
At the Intel SHIFT Conference, Amir shared the stage with several of our customers who are finding real applications for AI. Dr. Rick Lemoine at Sharp Health Care explained how his team used AI to sift through years of patient data and develop a model that identifies with 80 percent accuracy patients who will need an emergency response within the next hour. And Bryan Barber, a VP at Warner Bros., is kicking off an AI project to market and catalog home video content with the goal of increasing audience engagement – and ad revenue.
Amazon Web Services (AWS) is deploying Intel® Xeon® Scalable processors, optimized for deep learning engines to speed up inference performance by 100x and open the door of AI to businesses anywhere. Dan Neault, Director at Amazon Web Services said their machine learning stack enables organizations like Zillow to retrain 7.5 million models every day to provide the most accurate estimates on over 110 million home values. And, Netflix is able to make individualized predictive recommendations for its 100 million subscribers across 25 petabytes of information.
AI is changing what’s possible. And like mobility, social media, and the Web itself, it’s raising the bar for the competition. So you must begin to figure out what it can change for you. To get you started, Amir posed some questions for the audience to consider:
- In what specific areas could I use AI to boost my top or bottom line and help transform my business?
- How could we benefit from the better decision making AI enables?
- Who in our organization could provide the business and technological leadership to move us ahead?
- Do we have the right data to enable a transformation with AI?
- What hardware and software infrastructure would we need to invest in to build and deploy AI at scale?
Visit our Web site to learn more about how Intel is helping turn AI into a strategic advantage for a wide range of organizations.
 "Just one autonomous car will use 4,000 GB of data/day", Patrick Nelson, NetworkWorld, Dec. 2016.
 Cisco, “Cisco Global Cloud Index: Forecast and Methodology, 2015–2020,” 2016.