Explainable AI means humans can understand the path an IT system took to make a decision. Let’s break down this concept in plain English – and explore why it matters so much.
SAS CIO: Be proactive in leading your organization into the sensor-enabled world
For Christmas last year, I gave everybody on my list an Amazon Echo. They were a hit. My 11-year-old nephew delighted in pitting Echo’s Alexa against Apple’s Siri to see which device answered questions like, “What is 100 quintillion + 900 quintillion?”
As I watched Echo Dots turn into the new conference give-away, like iPods and iPhones a few years back, it became clear the year of streaming insights has arrived – and not just for consumers. Businesses have awakened to the potential of the sensor-enabled, real-time world described as IoT. Eighteen months ago people said, “When I hear about IoT, I just don’t see how it can apply to my business.”
That conversation has changed.
Now the discussion starts with, “I completely get that I need to look for places to enable steaming data to transform my business.” IoT is here to stay, and CIOs are scrambling to answer the question, “Where is the business value?”
The value of IoT lies in deriving new insights that improve or eliminate existing processes. The increased adoption of sensors means we’ll have more data coming at us faster than ever. This will drive businesses toward smart systems and an inevitable need for predictive analytics nimble enough to hang with ever-flowing data. As 2017 unfolds, one of the big transformations we’ll see is the adoption of streaming as the way to ingest data in real time. Doing so will transform data management.
Embrace analytics in flight
Today, IT professionals are familiar with ETL, which stands for Extract, Transform, and Load. It describes specific data management processes: Extracting the data and landing it into a data warehouse; cleansing it and transforming it; and loading it to do work. To get value from this new volume of data, we must eschew the old processes to keep pace. Instead of capturing data, putting it on disk, doing basic activities to clean it up and transform it, we’ll need to do it all on the fly. Instead of ETL, think in terms of Transform on Load (ToL). ToL is the ability to apply data transforms and derive analytic insights in the stream – “analytics in flight,” if you will.
In cybersecurity, for example, analytics in flight is paramount. We collect data from the network then use an identity management system to attach identity attributes in the stream: Who is associated with a device’s IP connection, and what are that user’s rights and privileges? Then we use the configuration management database to attach asset attributes in the stream: Is the device assigned to HR or finance? We’re able to immediately load the transformed and enriched network transaction into memory and begin analysis, looking for outliers and potential threats. Using traditional ETL data warehousing techniques would mean finding the threat days or weeks too late.
This shift represents a massive data management processing transformation as we continue the march to sensor-enable our world. This level of change will affect IT organizations with regard to the speed and agility of our systems. For example, streaming data enables immediate analysis and response. This concept comes to life with self-driving cars. They have sensors that do it all from understanding traffic patterns to detecting blind spots to automatically parking the vehicle. Waiting for traditional data processing is too slow for this level of analysis and response.
Be a leader
While there’s still a hint of hype surrounding the IoT discussion, it doesn’t mean the opportunity IoT represents isn’t real. Business unit staff members are technically savvy today, and they’re inundated with examples of technology innovation. They’re bound to see some of those innovations as an opportunity. To a certain extent, all lines of business are being transformed into software businesses because technology is playing a bigger role in how they serve their customers. As a result, there’s an increasing opportunity – even a requirement – for CIOs and their IT teams to evaluate transformative technology more quickly. Sooner or later, every CIO will hear questions like, “Why can’t I do this with sensors?” Be prepared.
Acting now means you still can be a leader in your space regardless of your industry. See how leaders in a few industries have put sensors to work, and imagine how you can replicate those successes on your own terms.
Retail. Retailers employ IoT through beacons and geofences to understand customer movement and preferences as they move through physical locations. Brands collect data from beacons. If a consumer opts in from a mobile device, a retailer can see activity and movement, and that opens the door to targeted, relevant marketing messages. When geofences are set around physical bricks-and-mortar locations, marketers can push offers to mobile apps once an opted-in customer comes inside.
Manufacturing. Manufacturing represents the industry where you hear about IoT and connected devices the most. It’s accepted that cars and trucks, airplanes, buses, and trains – essentially any transportation mode – operate with embedded computers and sensors emitting a constant stream of data to various smart locations. Manufacturers are now applying sensors to consumer electronics and appliances. Manufacturers use this data to improve the quality and design of the goods they create, and to market directly to consumers. In fact, manufacturers like John Deere, Ford, and Caterpillar that once relied on dealerships and agents to sell their products are moving toward direct sales models enabled by sensors to engage directly with the end consumer.
Insurance. Data collected by manufacturers – especially auto manufactures – can be used by other industries such as insurance. For example, telematics data collected from vehicles from on-board devices provide data on driving behavior, vehicle performance, and potential system failures. Telematics data aggregators and providers sell this data to insurance companies, and insurers use the information to inform policy underwriting and market to consumers. Imagine receiving an email from your insurance company informing you that your teenaged daughter drives too fast in the family minivan. That behavior may mean you need to increase your coverage on that vehicle to protect yourself.
Telecom. In telecommunications, streaming analytics helps assess valuable data about a consumer. It's becoming common for internet service providers (ISPs) to gather data about how their customers engage with the network. For telecom, the network is the sensor. They can monitor how much of a subscriber’s data plan is being used along with the frequency and speed of consumption. Gathering this information and stitching it together with other data – demographic and census information, for example – allows the ISP to make better system expansion plans.
Financial services. With the traditional bank branch system waning, financial services institutions must collect data from other channels like branded websites, mobile apps, and contact centers. For example, more data is being streamed from ATMs and mobile banking apps to inform banks about how to dynamically place and serve content to customers through digital channels.
Healthcare. Sensors allow health providers and insurance companies to collect and analyze patient data that can improve health while driving down medical costs. Wireless sensors embedded in wearable devices make it easier for a physician to track a cardiac patient’s weight, pulse oxygen level, and respiration rate. These vital signs can indicate deteriorating health, and detecting changes earlier can trigger life-saving interventions.
With sensors becoming more prevalent and affordable, it’s the CIO’s job to step in and triage the value and the opportunity presented. CIOs need to be the ones to saying, “Don’t just focus on the sensors. Think about the strategic value they can create for our business.” Ideally, the CIO has a technology-capable organization that understands how to apply technologies to the problems presented by business units so they can cross the chasm to a solution together.
Amazon’s Echo may represent the latest and greatest in household IoT, but it’s CIOs who must represent – and sort out – the greatest opportunities for sensors in their businesses today and tomorrow.
And, by the way, my nephew discovered that Echo’s Alexa gave much better answers to math questions than Siri or Google Home. Just thought you’d want to know.