Double takes on AI and 5G: Over-hyped or under-reported?
There are two versions of AI and 5G. One is over-hyped, the other is under reported. History tells us that over-hyped technologies were long on promises and short on delivery, but right now there is a more significant issue.
For AI and 5G the way these terminologies are covered on the Net is obscuring the business case for the deliverable versions, says freelance technology writer, Bob Emmerson. Moreover, in addition to providing pragmatic benefits today, the experiences gained should enable a smooth transition to the time when those promises are being realised.
Narrow and General AI
Let’s start with Artificial Intelligence, a scary, confusing topic that has been widely debated, one that’s ended up meaning different things to different people. For example, yesterday’s data mining solutions are being rebranded as AI. Therefore let’s make a pragmatic start and define intelligence as the ability to acquire and apply knowledge and skills and then define AI as being the business of getting computers to do tasks that would normally require human intelligence.
So far so good, but before covering those tasks we should underline the fact that there are two versions: Narrow and General AI. Narrow AI is where we are right now. General AI is where the Net tells us we are going.
Narrow AI (aka weak AI, an unfortunate term) automates the boring, repetitive parts of many jobs and lets people take care of the parts that require care and attention. It focuses on specific tasks that the technology performs much better than humans. It can be a relatively simple task, which it does 24/7, or when given a deep learning algorithm it can be demanding.
Given enough pictures of skin cancer an AI system can become better and a lot quicker than experienced doctors in spotting the disease. But does this represent real intelligence? There is no thought process. The system, the machine, merely leans from the pictures, from being instructed; these pictures indicate cancer, those don’t. It can perform this specific task 24/7, but the system doesn’t know what cancer represents. It’s a machine learning app and machine learning is a ground-breaking development that doesn’t need rebranding as AI.
General AI (aka strong AI) is expected to be able to reason, solve problems, make informed judgments, plan, learn, leverage prior knowledge and be innovative, imaginative and creative. That is a very big ask and it gets much bigger when you add the fact that for systems to realise real human-like intelligence the ability to experience consciousness would be needed. That’s a tall order, and it’s debatable if it can be realised in the foreseeable future, which is reassuring.
5G NR and 5G NSA
The two versions of 5G are 5G NR (aka Standalone 5G) and 5G NSA (Non-Standalone). They’re not exactly catchy terms and without an explanation Standing or Not Standing are meaningless words. It’s not clear which one is better.
Standalone is the over-hyped version, which will deliver three generic network services: mMTC, URLLC and eMBB; typical telecom-insider terms. Deployments will start with eMBB (enhanced Mobile Broadband); it’s the service that operators need to accommodate the big numbers in the consumer sector but hold your breath for a few years.
5G NSA uses the LTE radio access and the core network technology that is employed in 4G networks. This combination enables operators to provide 5G-type services over LTE that have more bandwidth, throughput, capacity as well as better connection reliability. Moreover these services can be delivered in much shorter times and at less cost.
The enhanced LTE performance comes from NSA’s addition of 5G base stations, dual connectivity and dynamic spectrum sharing. Dual connectivity with LTE and 5G radios allows devices to send and receive data using LTE and 5G simultaneously. This means that the network is carrying traffic over both 4G and 5G, each in its own spectrum. Therefore connectivity is much faster.
There’s a lot to like about 5G NSA, but like Narrow AI, it’s under-hyped. That was my double take.
The author is freelance technology writer, Bob Emmerson.