Certain IoT roadblocks will prevent companies from building intelligent processes. What will it take to clear them away?
In this Fourth Industrial Revolution, the Internet of Things (IoT) and artificial intelligence (AI) can help companies increase productivity and growth. But two barriers – incomplete data and wrongly calibrated sensors – stand in the way of achieving benefits at scale. As these barriers fall away, products and services will become far more personalized. Success in this new arena will require companies and governments to build strong trust-based relationships with those they serve.
Here, Now and Beyond - Intelligence realized through IoT
The IoT is the heartbeat of the Fourth Industrial Revolution – a revolution of game-changing innovation made possible by the combination of big data, analytics and physical technology. The volume of data that new web-connected systems will have available, combined with their ability to self-enhance through increasingly sophisticated artificial intelligence (AI), could fundamentally change how society operates.
Businesses have two primary goals from this IoT and AI-powered revolution: productivity and growth. Despite efforts to improve efficiency and offset cost through labor arbitrage, businesses need new approaches to offset years of productivity decline. The use of emerging technologies to measure such things as asset utilization, combined with AI and real-time decision management systems give many businesses great hope of increasing productivity.
IoT, AI, data management and cloud solutions also give businesses great hope of accelerating growth. Companies can build platforms to listen to customers, recognize their behaviors and preferences, and better address their needs – even to the point of predicting what they want – and that leads to growth.
Indeed, the Fourth Industrial Revolution gives companies the chance to move from dreaming about growth and productivity improvements to making significant strides on both fronts. But two barriers stand in the way of achieving the benefits of IoT and AI at scale – barriers that we can help tech vendors remove, before they become insurmountable roadblocks.
Incomplete data
AI is dependent on data which, in the Fourth Industrial Revolution, comes in large part from the IoT. In entirely new environments, such as smart buildings or self-driving cars, that contain huge volumes of sensors and are supported by powerful computing, everything imaginable can be smart and connected. But in most companies that’s not the case. A significant share of the systems running manufacturing shop floors, corporate offices and transportation hubs across Western economies is decades old. So, when companies want to implement intelligent automation to drive productivity or revenue growth they have to recognize the strengths and voids in the data driving those efforts.
It is rare that all parts of a process are automated. There may be human intervention and human reporting of key measurements in some parts of the process and areas where digital information trails are supplemented by paper reports. Those gaps, or parts of the process that are not measured by sensors and automated reporting, create challenges when trying to implement AI and other tools to drive a fully automated system or smart infrastructure.
To fill these data voids, we developed Intelligent Process Optimization (IPO), a capability to predict what a reading should be when a physical sensor does not exist. The concept is based on sophisticated neural network models used to predict what a measure would be if measured physically. Filling these gaps through “soft sensors” helps organizations build bridges across a partially automated process so that AI has the data it needs to read, react to and predict patterns across a process.
Falsified data due to wrongly calibrated sensors
When people watch the oven while cookies bake, and they see or smell something that seems wrong, they intervene. But in an automated world, rather than a person overseeing the process it’s a robot. The robot uses IoT sensors combined with AI to monitor and determine when an intervention is needed and to take the necessary action.
The sensors informing the robot have to be calibrated to send accurate signals, otherwise the robot will misread the situation. It will have a false view of reality and no experience to recognize that. In the baking example, if a sensor is indicating the temperature is correct when in fact it’s too hot, the robot will not intervene and the cookies will burn.
This same analogy applies to the use of IoT and robots at scale. If an IoT sensor is wrongly calibrated, small errors going into the process can produce disastrous outcomes. That is why, as use of IoT and AI grows, calibration of IoT devices will become both critical and challenging.
In a single manufacturing shop floor where there may be at most a few hundred sensors, calibration can be done by humans. But when there are millions or billions of devices in a smart city, for example, humans cannot feasibly do the calibration. Recognizing this significant barrier to scale, we are developing a Calibration-as-a-Service platform that uses AI on IoT sensor networks to calibrate, warn and correct sensors in real time. That way companies can be sure the data they use for processing is accurate and reliable.
Product/Services of the future: A future of Trust by Design
As companies move aggressively through the integration of emerging technologies, they can achieve their productivity and growth goals with companies like SAP and EY, who are thinking ahead about the challenges their clients may face as they scale. But companies must also think beyond the Fourth Industrial Revolution to the Fifth. We believe the next evolution will be defined by products and services ¬that are far more personalized than what is available today.
For example, in the current revolution, an insulin pump can automatically measure a person’s blood sugar and notify that person about the amount of insulin he or she needs based on their weight. But one can imagine a product that learns by an individual’s behavior, amount of sleep, and other factors and correlates that with the person’s condition prior to recommending or injecting specific insulin levels.
Similarly, what if a music curation tool could not only consider one’s musical preferences but also other factors. In the Fifth Industrial Revolution such a service may use sensors to measure mood and offer active, romantic or other options based on a measure of one’s whole condition.
What is needed to create this level of personalization? Users need to trust not just the product but also the company, the way the product is created and how its ecosystem of data is used. Trust in companies and governments will be the most critical element in the future for smart, integrated, personalized products and services.
