5g, AI, and the mobile revolution

We call our phones “smart,” but how intelligent are they? Anyone who has tried to use their handheld device to access audio streaming, or composed a message using voice recognition knows the answer sometimes is, “not very.”

The fact is, our phones can do certain things really well—perform searches, take photos, help us surf our social media accounts, play music—and there is so much more they are working to improve on.

When we travel to a new town, our phones cannot suggest a vegetarian restaurant for dinner, and then make the reservation for us without being prompted.

They cannot weed through the many possibilities presented on our dating sites automatically and alert us to one particularly good match.

They do not call an ambulance if we suddenly need help and tell the medics what went wrong.

The main reason our phones are not truly intelligent yet is because they lack key qualities our minds possess—especially memory and processing speed.

The human brain can store around a quadrillion bytes of information, and can process entire images in as few as thirteen milliseconds, research shows. Our phones—even those with the most memory available today—fall short in comparison.

Without processing at the speed of thought or oceans of memory from which to glean insights, our mobile phones are limited, and are mere machines executing our commands. What is so smart about that?

The 5G superhighway

A new day is dawning, however, and sooner than most of us realize.

The advent of 5G cellular technology looming on the near horizon promises to open up a slew of new lanes on the information superhighway, reducing or even eliminating traffic jams, allowing vast quantities of data to move faster and more freely than ever before.

Enabled by innovations in memory and storage, 5G, coupled with next-generation technologies including artificial intelligence (AI), offers enormous promise for transforming our phones into prescient, intuitive, truly smart companions.

Given the amount and extent of mobile phone use today—the largest computing platform in the world and soon to be our only computers, some predict—the 5G/AI revolution stands to change the way we communicate, the way we work, the way we consume and recreate and experience.

Unlike previous advances in wireless technology, the advent of 5G constitutes much more than a step change: it’s a giant leap in data-transmission speeds, responsiveness, and connectivity. Each “G,” or generation, has brought significant changes in our phones’ abilities, but 5G will utterly transform our devices and our daily lives.

Here’s a snapshot of how we got to this exciting place:

• 1981: 1G. Analog cellular technology for voice-only offers no on-device data storage or processing, and transmission speeds of 2.4 kilobits per second (kbps) or less per second. Phones are very limited in their range, and signals are highly susceptible to interference.
• 1991: 2G. The technology switches to digital for voice communication, and also accommodates text messaging, but phones still lack data storage or processing. Data transmission speeds increase to 40 kilobits per second (kbps)—still very slow by today’s standards.
• 1998: 3G. Now our phones can store data [32 Megabytes (MB) to 16 Gigabytes (GB)] and browse the internet. Data speeds increase to 21.6 Megabits per second (Mbps).
• 2008: 4G. Data optimization supports video streaming. Memory increases to 16GB-256GB storage, and transmission speeds ramp up to 1 Gigabit per second (Gbps).

Each of these advancements has been monumental in its own right, but in the end, the cellphone is still a device that does what we tell it to, and little else.

Starting in 2019, however, the advent of 5G will bring new capabilities to mobile devices which will expand the need for storage capacities of 512 GB in 2019 (and, by 2021, 1 terabyte), data transmission speeds of up to 20 Gbps, and bandwidths allowing connections to many devices at once including sensors and other “smart” devices such as autonomous vehicles.

These advances will enable the cell phone to move beyond its current role of passive servant to active participant, able to process vast quantities of data from a plethora of sources in real time—response times, or “latency,” will shrink—to affect our decisions, experiences, and lives in ways we can only imagine.

It will achieve this using a number of technologies, including artificial intelligence (AI).

Minds of their own

The fast processing speed, low latency, high bandwidth, and vast storage capabilities 5G wireless technology offers will at last bring to our mobile devices intelligence to rival the human brain.

For the first time, we will hold in our hands a responsive, pro-active personal assistant able to understand us and our desires in the context of our preferences, environments, situations, and unique personal attributes and limitations, and to guide our choices and experiences accordingly. These capabilities are at the heart of AI.

Think of what, in its infancy, AI can already do: recognize faces on social media (and, in some cases, on our phones); transcribe speech and translate languages; find cancer in blood samples better than humans can do; detect impending breakdowns in manufacturing equipment; fly drones and drive cars, and much more.

Many of these applications, however, depend on programmed algorithms. 5G’s expanded capacities will allow a barrage of unstructured information via video and audio streaming, still imagery, text, and code from humans as well as from objects communicating not only with us, but also with one another.

Expanded storage, bandwidth, and 5G will enable new features in our phones and edge technology that previously were not possible.They will be able to send information from smart bandages to our doctors about how our wounds are healing. They will be able to stream video games and let us play them remotely with others.

Add AI into the mix, and our devices will be able to process all the data required for these tasks instantaneously and adjust its responses according to not only the current context but past lessons, as well. Being able to learn will make our devices truly intelligent—to advise us to apply an antibiotic to that wound, for instance, or to change the bandage.

In factories, AI’s machine learning capabilities may enable factories to run themselves as cyber-physical systems, with little human intervention. Communicating with one another, machines will monitor production, correct errors as soon as—or before—they occur, replace wearing parts, scale up or down production according to real-time demand, and more.

AI will also empower autonomous cars to communicate not only with us via 5G, but also with other vehicles and with sensors on streets and highways to change course, slow down or speed up, and perform the countless other calculations and actions that humans conduct while driving, often without conscious thought. It also will enable computing devices to manage traffic.

It’s all in the chip

The advent of 5G will change not only the capabilities of our mobile devices, but also their memory and processing requirements: for AI applications, especially, they will need more memory and much faster speeds than they now have. As a result, today’s AI applications depend on servers located in the cloud, which depend on internet connections to send and receive data. All this back-and-forth takes time, slowing processing speeds and creating data “bottlenecks.” A slow phone is hardly a smart one.

But 5G’s expanded capabilities will remove network bottlenecks by making it easier for our phones to communicate seamlessly with the cloud and also retrieve data stored on the device (“edge” computing) in real time—at the speed of human thought, or even faster. (Not incidentally, edge computing is also more secure, and saves energy.) To do so, however, our phones are going to need to read and write data a lot more rapidly than they can do today.

The bigger information “pipeline” that is 5G will accommodate much more data at one time, as well. To keep all that information readily available, phones will need a vastly expanded storage capacity. Memory and processing are where Micron comes in.

Many of today’s mobile phones use a flat, 2D flash memory storage chip known as NAND. 2D NAND has served us well in the age of 4G, but as we use our handheld devices for more and more tasks, this form of memory will soon no longer be able to keep pace with our needs.  

Anticipating this problem, Micron has developed fast and efficient 3D NAND, which stacks layers of data storage cells, taking up the same amount of space as a single NAND chip but tripling storage capacity.

To move all that data quickly through the expanded 5G bandwidth, our devices will need lots of processing power. Micron developed our powerful LPDRAM memory to transport high volumes of data through computing systems at very fast speeds—essential to the low latency that AI needs to work properly. But this can be done with low power (the LP in the DRAM) requirements, to help extend battery life.

Micron technologies in development will combine the best of 3D NAND—high data storage capacity—with the most desirable features of DRAM—superfast processing—for intelligence that can only be beat by the human brain. For now.

“Data is the key enabler of AI,” Micron senior fellow Mark Helm explains. “If you compare the workloads of computing systems that are executing historical workloads versus AI workloads, data is a much more important factor for AI.

“That raises an opportunity for Micron to be the keeper of the data, whether it be very low latency access via DRAM or very high capacity storage from a NAND flash system. Our control over the data gives us a great opportunity to leverage value out of AI in the future.”

Micron is committed to exploring AI’s potential in health, infrastructure, energy and other fields. Visit micron.com and join us on Twitter, Facebook and LinkedIn for more information.