What is embedded intelligence and how can CIOs prepare for it?

Advances in AI hardware and software are pushing artificial intelligence and machine learning capabilities out of the data center and into a company's products.

New research from Deloitte Consultancy LLC, "Pervasive intelligence: Smart machines everywhere," suggests that embedded intelligence will affect just about every industry and could help companies save money as well as expand into new markets.

In this interview, David Schatsky, managing director at Deloitte and a co-author of the report, also sounds a warning bell for CIOs: Keep your eye on the trend because embedded intelligence could be a disruptor.

Editor's note: This interview has been edited for brevity and clarity.

What is pervasive intelligence?

David Schatsky: Artificial intelligence is moving out of the data center and into the real world. It's appearing in factories, in the field and even in people's bodies. And because it is increasingly possible for objects and machines and devices to have AI embedded, we're going to be living in a world where AI is pervasive -- where we're surrounded by smart devices and machines that will change the way we live and work.

Give an example of this kind of embedded intelligence in action.

David Schatsky

Schatsky: The healthcare example is distinctive because it involves AI embedded in a person's body. So, this is the ultimate in embedded [intelligence]. It has the potential for delivering better health outcomes to individuals who are at the mercy of their own bodies. The idea of applying machine learning to biometric data that is continually gathered from a patient and using that [information] to be able to, say, reliably predict when a seizure might be coming and then to take action to prevent it is a pretty awesome application, I think, in terms of well-being.

How has the hardware and software advanced to enable pervasive intelligence?

Schatsky: Among the biggest is the development of processors or chips that have the power to run AI algorithms while consuming little space and little energy. AI calculations have been a very computationally intensive, energy-intensive thing, and that's why most of the work has been done in giant data centers with giant machines. But the hardware is getting better and better in supporting AI types of processing in low-powered machines. That's one of the key enabling technologies here. You don't need to go over the network to the cloud to perform a calculation. You have everything you need right on a machine or on a device -- even if it's not connected to a network.

What's making those processors less computationally intensive?

Schatsky: Part of it is the makers of chips that go into mobile phones, the makers of new processors that are coming out are intended for things like embedding in sensors and small pieces of equipment. These processors are optimized for performing the kinds of calculations used in machine learning. And because they're designed with that in mind, they can perform these processes more efficiently, with less power consumption. So, that's a big part of it. Mobile phones increasingly have these kinds of processors embedded in them; there are dozens of startups working on processors that are designed with machine learning in mind. And when you design a processor for specific tasks, you can make it more efficient at performing those tasks.

And what about the software?

Schatsky: In parallel, even as the hardware is getting optimized, because this idea of having embedded AI is such a promising capability, researchers have been working to optimize the algorithms themselves so that they can be more efficient -- make more efficient use of memory, for instance, and run more efficiently.

How can CIOs help prepare their companies for embedded intelligence?

Schatsky: These are new, enabling technologies -- embedded intelligence, embedded processes that have AI capabilities, embeddable algorithms. CIOs who work for product companies that could potentially benefit from this need to get a grip on the technology trends. They need to understand who the technology providers are and how they might impact the company's offerings. And I would think that they are going to play a key role in helping to advise on the architecture of new products that use these technologies.

Do you see embedded intelligence as a disruptor?

Schatsky: It's an opportunity and a threat. If you have a product or service that could be affected by a smart version of those products or smart products used to disrupt the service, then, yes, it's something you need to pay attention to. Either look at how to incorporate that technology in your own products or look at how it's going to change your business.