Artificial intelligence (AI) and machine learning (ML) can be very cool technologies. They will (and already do) play an enormous role in solving humanity’s biggest problems, extending our lives and improving their quality,
and unlocking some of the secrets of the universe.
The question of how AI is going to do all this is more difficult to answer. And sometimes, the people who develop AI can’t answer that either. Take this story from The New Yorker, for example. It’s about an AI developer who built a system that identified different types of bread to streamline and sanitize checkout processes at bakeries. Today, that technology is used to identify cancer cells, detect problem bolts in jet engine parts, and measure tiny particles in physics experiments, just to name a few applications.
But here’s what we are seeing: in our industry and adjacent ones, plenty of folks are kicking the tires. They’re using AI and ML internally to improve operations and selling AI and ML technology to their customers to drive revenue. It’s helping IoT manufacturers and service providers improve R&D and manufacturing processes, and maximize their product life cycles. Cybersecurity professionals are using it to protect data. AI and ML is automating warehouses and helping us find the most qualified candidates for the job. Big and small, AI is solving problems and allowing us to do things that we could never do before.
What is AI and ML?
Many different technologies come with the label of either being AI or using AI and machine learning technologies to give you some sort of desired outcome. But just because some solutions provide the same outcome as AI (automated processes and data analytics), that does not mean that the solution is using AI or ML. There is a difference between explicitly instructing a computer to do something and programming it to learn and improve on how it does something. So before we explore who is doing what, let’s settle on what artificial intelligence and machine learning are.
Artificial Intelligence: an array of technologies that enable machines to sense, learn, understand, and act
on their own, based on training data and the AI’s experiences.
Machine Learning: a branch of AI focused on the use of algorithms that can improve automatically through experience. It enables computers to perform tasks without having to be explicitly programmed to do so.
Who is doing what with AI and ML?
There are many businesses using AI or creating products that interact with AI in one way or another. Or maybe they’re using AI to find out which product to make next, finding out better ways to make or service that product, or to do something that wasn’t possible before. But just so we can get a feel for the different AI solutions that are out there, let’s use a few specific examples of companies to illustrate.
COVID-19 created a logistics nightmare. Shortages (labor, raw materials, you name it, there is a shortage) coupled with an increased reliance on e-commerce has put an incredible strain on the entire supply chain. Manufacturers are having trouble getting all the components they need to manufacture their products, and are paying a premium when they do. On top of that, getting shipments from point A to point B is so slow that global shipping leaders have suspended their service guarantees.
The folks at DHL are leveraging AI to overcome their struggles. In September, the company announced DHLBot, an AI-powered robotic arm built by Dorabot that sorts parcels. According to the company, DHLBot can sort 1,000 small parcels per hour, increasing productivity by 40%. The powerful AI enables the robotic arm to quickly sort packages into separate delivery bins, creating efficiencies on the spot.
It has been a decade since IBM’s Watson defeated Brad Rutter and Ken Jennings on Jeopardy. Retired from the game show circuit with a perfect 1-0 record and $1 million in winnings, Watson has moved on to bigger and better things. IBM’s AI platform is used in everything from risk and compliance, finance, and IT operations, to customer service, healthcare, and video hosting and streaming.
In the security space, IBM QRadar Advisor with Watson provides cybersecurity professionals with a much-needed tool to deal with the overwhelming task of defending our data. The solution automates routine SOC tasks and connects the dots that humans cannot. This helps security analysts use their time more effectively. And considering that we’re in the midst of a cybersecurity labor shortage (with no end in sight), you can see how tools like these are going to benefit us all.
IBM’s AI for customer service technology has also played an integral role in helping some businesses enhance their customer service operations. The solution enables businesses to create chatbots that integrate with their CRM solution, respond to spoken requests, and listen in on conversations between customer service reps and customers, and provide agents with helpful information based on what’s said during the conversation. Not only does this enable businesses to enhance productivity in existing call centers, but it also provides them with the means to interact with customers like they never could before. It’s yet another example of how AI helps us break new boundaries.
The company claims that Watson users were able to achieve a 337% ROI over a three-year period.
The folks at Canon have a few AI and ML irons in the fire that are being used to make advances in the office, camera, and medical businesses.
In its office-facing businesses, Canon is using AI and ML to automate business processes, like invoice processing, claims processing, eDiscovery, and digitizing mailrooms. Most of the technology sits where paper is used in a mostly digital process, automating scan/capture processes, and thereafter making simple decisions that cannot be programmed (or would be impractical and incredibly time consuming for humans to program).
In the camera business, Canon is implementing AI in both its consumer- and business-facing products. On the consumer-facing side, there is the PowerShot Pick. The device can automatically detect a person’s face — even their facial expressions — and “knows” when it’s time to take a picture. Meanwhile, on the business-facing side, Canon’s $2.8 billion investment, Axis Communications, is leveraging AI and deep learning in its security cameras. For example, the AXIS Q1615-LE Mk III Network Camera leverages Axis Object Analytics, which can distinguish one “object” from another (for example, bikes, cars, trucks, busses), decide which one requires attention, and take action if needed (let’s say, by issuing an alert). The device also works with third-party AI applications to handle additional processes.
In their medical business, Canon is leveraging artificial intelligent Clear-IQ Engine (AiCE) technology for CT scans and MRIs. MRI and CT images are fed into the respective solutions, and analyzed pixel by pixel, eliminating noise when detected, and creating a clearer image. With clearer images, healthcare providers can spot hard-to-find problems. The solution can also detect abnormalities or anomalies within a CT scan or MRI. This allows healthcare providers to perform exhaustive analysis on each patient’s image, while reducing the amount of time that healthcare workers spend manually analyzing images. All that saved time can be used to ensure that all patients get the best care possible. This technology will literally save lives.
Konica Minolta is making a similar AI play in the security camera market. In 2016, it purchased a majority stake in Mobotix, and used Mobotix’s technology in conjunction with its Dispatcher Phoenix Platform to help businesses return to work safely. These cameras replaced the need for a human to take another’s temperature, ensuring that temperature screenings can be carried out while maintaining proper distancing procedures. This application of the cameras in and of itself isn’t an example of AI, but rather an innovative way that Konica Minolta leveraged its technology to create new solutions when COVID-19 reared its head.
But these cameras, like Canon’s, do leverage artificial intelligence. Mobotix’s cameras can count how many people are in a room or determine whether someone is wearing a mask or not. More importantly, these kinds of applications allow us to do things that humans couldn’t reliably do (or do at all). A human, or even a team of humans, wouldn’t be able to reliably tell you how many people are in a specific space at any given time. Likewise, while spotting a maskless face in the crowd can be done, asking a human or group of humans to take on this task is not very effective. If there are 50 offenders among 1,000 in a room the size of a tradeshow exhibition hall, do you think you’d be able to spot them all? You might catch a few, but you probably won’t be anywhere near as reliable as an AI-powered camera would. This isn’t about saving time and money — this is about doing things you weren’t able to do in the past.
Konica Minolta is also leveraging camera-based AI in more specialized applications. The HitomeQ Care Support solution leverages sensors to keep tabs on patients in elder-care facilities. The sensors can detect when a resident has fallen out of the bed, for example, to dispatch a care worker immediately. KM’s AI can also be leveraged to monitor shoppers in a store, or to find defects in a box before it leaves a factory.
AI is everywhere. Many of the use cases are less flashy than IBM Watson winning Jeopardy, but there is plenty of money to be made by automating mundane processes (like monitoring security cameras, processing invoices, or answering a customer’s question), and enabling businesses to do things they’ve never done before.