Are friends electric?

Róisín Kiberd takes a look at the industries where AI is your co-worker

5th July, 2020
3
Róisín Kiberd takes a look at the industries where AI is your co-worker

“They learn to speak, write and do arithmetic. They have a phenomenal memory. If one read them the Encyclopedia Britannica they could repeat everything back in order, but they never think up anything original. They’d make fine university professors . . .”

The above is a line from RUR, or Rossum’s Universal Robots, the 1920 play by Karel Čapek which popularised the term ‘robot’, and at the same time raised questions around whether machines would usurp their human creators.

Today robots can speak, write and do arithmetic too. In 2019, reports even appeared of an English language professor in Germany whose assistant is a robot named Yuki. Whether they have original ideas, however, is open to debate.

What we do know is that at present, in industries as diverse as healthcare, manufacturing and construction, artificial intelligence (AI) and augmented reality (AR) are making a significant difference to everyday business, and can for you, too.

Automation doesn’t need to create a dystopian scenario, where robots replace human workers. At present, across numerous industries, AI functions in an advisory capacity or as an assistant, retrieving data and modelling future outcomes with a dynamic and versatile technique called the ‘digital twin’.

“We’re seeing AR show up in different places like engineering or immersive surgery, which is getting really interesting,” said Marc O’Regan, chief technology officer of Dell Technologies, Ireland.

“Things like digital twins are being used, and have been used here in Ireland for a few years now. I know of at least three major hospitals using AR as part of assisted surgery procedures. Doctors use the technology to perform operations. It’s a developing field, but it’s starting to emerge from the shadows.”

The ‘digital twin’ as a concept sounds a little unheimlich; it’s a replica of a living being, or thing, or place or system, designed to mirror the life of the original and to anticipate every outcome after intervention. It’s one of many potentially life-saving AI and AR technologies currently in use in the healthcare sector in Ireland and abroad.

O’Regan also referenced another project, a connected ambulance in which data on the patient is instantly retrieved, and their condition is monitored, predicted and even potentially treated using AI.

“One of the use cases we’ve examined was a cardiac patient. The test was that if a patient suffers from a heart attack in the back of an ambulance, how can technology help first responders to treat them?” he said.

The team created a system converging real-time data – metabolic data, telemetric data and data gleaned with image recognition, aka ‘machine vision’ technology – all of which is communicated to doctors before the patient arrives at a hospital.

“The ambulance might be travelling through a remote, rural space,” O’Regan said, “but the data is being transmitted cell-to-cell across a network, and it’s hitting the hospital too, so it has detailed information on the patient before they arrive.”

AI can potentially save lives in this role, as assistant to doctors and first responders. The longer the technology is used, the more it learns and the more accurate its recommendations will be.

“It’s possible to accumulate data on the patient, along with tens of thousands of other patients, and assemble a dataset, then extract information from it and correlate the data in real time,” said O’Regan.

AR is also widely used in the fields of construction and architecture, where buildings designed in virtual reality are assembled in real life by headset-wearing builders and engineers.

“You can see a version of the structure of a piping system, or a roof, for example,” said O’Regan. “The coordinates and the blueprints are knitted into the AR package, and you’re able to see the blueprint as a whole.”

Buildings, too, can have a ‘digital twin’; in this case, the simulation can run different changes to the environment and anticipate how they’ll turn out.

“We’re making real headway in terms of bringing value, taking something from the idea stage to testing it in an environment that’s simulated, bringing it through phases and into production,” O’Regan said. “You can accelerate the stages of testing and development in a safe environment, and get it into production.”

A dizzying vision

It’s been predicted that the global 5G rollout will speed up AI adoption, allowing edge computing and Internet of Things (IoT) devices to make use of AI. IoT sensors will proliferate, creating more real-time data, which will in turn help AI to improve. It’s a dizzying vision of the future, one that’s not even particularly far away.

“We have five pillars in the office of the chief technology officer at Dell Technologies,” O’Regan said. “There’s edge, security, cloud-native applications, data management – and then there’s 5G, with 5G acting as an enabler for all the other pillars.”

Dell views 5G as far more than a wi-fi update: “This isn’t just about better broadband. It enables us to look at our AI models in so many new ways, land those models as close to the data source as possible, and enable them to make accurate calculations. They’re quick, they’re accurate, and they make a difference.”

Manufacturing is another area experiencing dramatic change due to automation. Samantha Cummins-Byrne is Irish development lead with Universal Robots, a Danish manufacturer of cobots, or collaborative robots, distributed in Ireland by Cobots.ie and Reliance Automation.

Cobots are six-jointed robotic arms which work safely and efficiently alongside human staff. “Our goal at Universal Robots is to improve the efficiency of businesses, increasing profitability and improving employee safety,” said Cummins-Byrne.

Cobots are easy and fast to set up, and can give organisations an edge without compromising safety or quality of product.

“These are the primary concerns manufacturing businesses are facing and actively trouble-shooting,” said Cummins-Byrne. “We know from experience with manufacturers that issues such as labour shortages, employee safety and even floor space can prevent them from competing in this highly challenging market.”

Factories and warehouses, and their employees, have suffered during the pandemic as companies struggled to operate while keeping employees safe. Automation might not seem like an option for small and medium-sized organisations, but Universal Robots cobots can adapt to environments on any scale.

“Our cobots can be situated in small spaces, and are intentionally built to interact and collaborate with humans,” said Cummins-Byrne.

The prospect of upskilling raises another question: how much expertise does the average automated product require? Will it require hiring someone specifically to manage these systems?

Cummins-Byrne said that most companies working with cobots in Ireland at present have designated automation teams, or leaders who head up the integration process, but that the training process is kept deliberately simple.

“It’s so easy to use a cobot. One person can complete our training and easily share their knowledge and skills with their colleagues. It’s great to work with a dedicated automation team as we can really get a feel for how the integration is going, and how the cobot is performing – not just how it’s functioning, but how it’s improving operations,” she said.

Neural nets – algorithms which replicate the biological structure of the brain – are currently at work in industries like manufacturing, aerospace, telecommunications and the automotive industry.

They sound like a futuristic technology, but they’re also of the past; the idea for neural nets was first proposed in 1944 at the University of Chicago. The concept was researched for almost a decade before it fell out of favour, only to be resurrected again in recent times.

“A neural network will take as much data as you can possibly throw at it,” O’Regan said. “The complexity of neural nets is absolutely unbelievable. They’re vast, absolutely vast. It’s like looking down on a picture of the ocean, and each pixel in that picture represents a different neural net. Each one of those neurons is making an independent calculation, and the network sitting under that is networking the different calculations.”

These networks will play an important role in the future, and the hardware which supports them will become smaller, more efficient and more durable with each passing year.

“Right now we have processors, coprocessors, GPUs and FPGA that we build into compute servers, which do the calculations for us. But we want to take all that and make it the size of a thumbnail,” said O’Regan.

“Then we want to make it resistant to heat and resistant to water. It needs to be able to carry extraordinarily complex models, built with AI. I think that’s where AI is going.”

The near future will see AI, AR and automation in general become more widespread, and more accessible to businesses of every size. Neural networks, in particular, present a dazzling possibility; a ‘digital twin’ to that most mysterious and unpredictable of things, the human brain.

Real life, and real technology, looks to become stranger than fiction. To return to the words of Karel Čapek, “Nothing is stranger to man than his own image.”

Share this post

Related Stories

Security Watch: Ransomware epidemic rages on

AI: the promise of better health through knowledge

Connected Jason Walsh 1 week ago

Smart infrastructure: the data things are made of

Connected Jason Walsh 1 week ago

DataSolutions sets ambitious carbon-neutral target

Connected Emmet Ryan 1 week ago