Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

The internet of things without AI-enabled IoT? Still dumb

While IoT-connected devices began in fits and started several years ago, today they’re gathering momentum.

That’s because the internet of things offers immense possibility by connecting hitherto “dumb” devices to the internet to provide intelligence and decision-making capability in the moment. These sensors could send data on products’ operational state and location as well as a host of other key data to help companies gather insight to make business decisions.

IoT devices lay the foundation for so many efficiencies and innovations today, from connected cars and smart refrigerators that tell us how to get where we need to go to sensors that can alert us about product malfunctions to healthcare wearables and mobile apps that can guide our health regimen to the supply chain, where IoT devices can track a shipment start to finish.

But over the past several years, while connected devices offered insight, the data volumes generated were fast and furious, and enterprises’ ability to structure their insights into forward-looking behavior remained incremental. With machine learning and artificial intelligence maturing at exponential rates and combining with IoT devices, true business insight is on the way.

AI-enabled IoT may finally yield the speed, insight and scale that IoT needs to flourish in enterprises.

“The reality is we’re moving from connectivity to intelligence,” said Anthony Passemard, head of product management for Google’s Cloud and IoT, speaking at Google Cloud Next 2018 in late July. “Without actionable insights on the data, it’s hard to get return on investment. Intelligence is key to those investments.”

AI-enabled IoT

Harnessing that intelligence has continued to be a challenge, though. By 2025, IDC predicts that there will be over 80 billion connected things creating and replicating more than 180 zettabytes of data every year.

But the speed and accuracy of IoT data needs to improve to be ready for the enterprise given the massive influx of data. With AI-enabled IoT, companies can whittle down the billions of data points they have into truly meaningful kernels.

Aker BP, an oil and gas company in Norway with some 2,000 employees, produces nearly 150,000 barrels of oil a day. It uses IoT to monitor its equipment, to protect its workers from harm and to reduce costs.

The company can pull data quickly, then turn it into meaningful action through AI-enabled IoT activities.

“One of the thing we see quite good results from, we are pulling up to 1 million data points per second into a data store,” said Kjartan Nesse, SVP of operations at Aker BP. “Based on this data, once we get it contextualized in the right way and push it back to the operators, that really helps drive decisions.”

Further, Nesse said, AI-enabled IoT data can provide predictive maintenance insights for equipment or provide insight on conditions in environments and “drives opportunities to move people out of dangerous zones.”

Companies are also using artificial intelligence and IoT to digitally recreate what’s happening in the real world. Freight Farms, which creates year-round agricultural environments globally, uses environments outfitted with IoT sensors to genera ideal farming conditions, such as soil, air and Co2 levels.

“We’re looking to lower the barrier of entry [to get into the farming industry] and make food supply a loT more accessible,” said Jon Friedman, cofounder of Freight Farms. With IoT, it’s possible to use data to create consistency and to optimize farming conditions.

“IoT can set the environment for a certain crop,” Friedman explained. “You can create the perfect day of summer in that environment. You can take the nutrients of Italy, pair that with air quality of Salinas Valley and combine that with the Co2 near an active volcano, and give a plant the light spectrum it wants.”

In this way, Freight Farms isn’t just mimicking real-world conditions through IoT. Rather, it’s creating an ideal, other-worldly set of optimized conditions. With a mobile app, farmers can optimize conditions beyond the barriers of the physical world.

“IoT is central to build these environmental recipes to match what is out there in this world, but also what isn’t available in this world,” Friedman said.

AI and IoT devices push enterprises to the edge

But the proliferation of devices and data making round trips to the cloud only floods this infrastructure with volumes of data it necessarily handle.

That’s why IoT providers — indeed IT infrastructure providers of all stripes — are moving compute-intensive processes such as AI-enabled IoT processes to the edge — edge computing, to be exact.

“All the things that are producing data, and the people interacting with each other and with things, it’s going to push data to the edge,” said Gartner’s Thomas Bittman at a conference on infrastructure in late 2017. “We can’t have enough pipes cheap enough to accommodate the amount of data out there.”

Accordingly, Gartner estimates that, as data is pushed to the edge, enterprise adoption of the edge computing will follow suit. “While today some 10% of enterprise-generated data is processed outside a traditional centralized data center or cloud, by 2022, Gartner predicts this figure will reach 50%,” the research firm estimated in “What edge computing means for infrastructure and operations leaders.”

Consider data-intensive capabilities such as artificial reality-enabled games, using videoconferencing in meetings or asking queries of voice-based digital assistants. All these kinds of data-intensive processes have to execute tasks in fractions of a second and call on massive amounts of data. These kinds of tasks are best executed at the edge, where devices can access resources without having to call on the cloud. Further, these edge computing architectures are better suited to industries with compliance requirements that prohibit their data from being sent to the cloud.

This convergence of IoT, AI and edge computing might be known as the intelligent edge, with AI-enabled connected devices that don’t rely on centralized architecture. It’s bringing the data insight, the compute resources, and the users and devices right where they need to be to take intelligent action in real time. Experts say this intelligent edge is inevitable given our current data and device proliferation.

“In the next few years,” Bittman predicted at the conference, “you will have edge strategies — you’ll have to.”

All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.

Data Center
Data Management