In the past couple of months, I have been approached countless times by individuals seeking my perspective on AI, owing to ChatGPT's ability to provide a human interface to this technology. Not just colleagues, but friends, family members, even complete strangers. They want to know if it’s as great as it’s hyped up to be by many people (like AI researchers, influencers and even academia), and as threatening as the media makes it out to be. You’ve probably been in the same boat. So, let’s get down to brass tacks.
Not all AI is created equal, so it’s not fair to say all AI is bad, scary, threatening or whatever other adjective is being used to stoke fear among the public and spur politicians to regulate the technology. Yes, some of the latest generative AI solutions have problems that must be ironed out, such as those that can throw up results that are inaccurate and/or biased. And there’s real concern that some generative AI tools can infringe on copyrights (e.g., ChatGPT). But, like any type of technology, these issues aren’t universal. So, while AI researchers and other technologists work with government leaders to address these very specific concerns about the behavior of certain AI, I think it’s important to emphasize that not all AI is created equal. AI, like people, can be taught to do the right thing.
In fact, AI is doing a lot of good in the world today – in your world, your bubble. This includes the adaptive AI primarily leveraged in Zebra’s technology solutions, in which real-life data is used to refine the outputs over time.
Manufacturers and retailers use adaptive AI to forecast demand and plan inventory, especially consumer packaged goods (CPG) companies. AI-powered machine vision assists with detailed quality inspections of everything from medications to automotive and electronic parts to 2-liter soda bottles and frozen dinners coming off the manufacturing line. Autonomous mobile robots (AMRs) equipped with machine learning and computer vision work in support of people in factories, warehouses and distribution centers, helping ensure they have the raw materials, products and packages they need to get products made, packaged and loaded onto trailers on schedule.
Front-line workers responsible for delivering services and experiences we’ve come to expect – to demand – have to keep people in the building and online happy at the same time. They are essentially serving at least two customers at once. Have you ever tried to carry on two different conversations at once? It’s not easy and, eventually, you’re just going to throw your hands up in the air and say it’s impossible.
Well, the only reason many essential workers are able to serve so many of us at once is because there’s trustworthy adaptive AI working alongside them somewhere. It may be working within an app on a wearable or handheld mobile computer, in a business’s inventory, logistics and/or e-commerce information system, or even in the cloud tied to an industrial automation system. But wherever it lives, it’s well-trained to “listen” to what everyone wants, “see” what is happening in the world (the nuanced changes or trends that a human might miss) and assess all angles of every situation before ultimately telling front-line and back-end workers what they should do now and next to make everyone happy.
In other words, the AI you would use to make your business run better – the AI you need to hit your business targets this year – is not the generative AI that’s drawing ire or fueling speculative fear among government officials, tech industry leaders, academic communities and the general public.
The AI you need to pay attention to as a business or team leader is the one that can take all the data generated by physical Internet of Things (IoT) objects – think barcode labels/scanners, handheld mobile computers, tablets, RFID tags/readers wearables, machine vision cameras, robots, and environmental sensors – and make it make sense to you and your team. And there are a lot of responsible AI-powered technology tools that fit this bill, ranging from workforce management and task management software to visual inspection systems, loss prevention systems, and forecasting/planning platforms, with more being ethically designed and trained every day.
A few years ago, everyone believed data was gold. Then we (meaning society) learned that data isn’t valuable unless it’s clean, aggregated, and thoroughly analyzed in the context of a specific request, whether that was to help avert or solve a problem, confirm the best option, or make some other decision. No one has time to filter through a mess of information, much less the capacity to process it in a meaningful way. It’s why human-led analytics and decision-making are often flawed. Our brains can’t work the same way AI can. We may be able to sense something is happening, but we struggle to analyze it fast enough to be able to act at the right time. Most actions are the outcome of educated guesses resulting from a bunch of assumptions derived from a few extracted facts.
But if we learned anything during the pandemic, it’s that the world can’t afford to keep acting on “best guesses.” We need AI to help us analyze and contextualize scenarios – to definitively spell out the “who, why, what, where, and when,” to point out patterns we would have otherwise missed and show us everything happening in the world.
Retail and hospitality-related decisions can’t just be made based on footfall or past consumer behaviors. Weather, road traffic, community events, and macroeconomic events happening locally and halfway around the world might all be influential on inventory sourcing decisions or labor needs. But we won’t be able to know for sure without the help of AI.
A manufacturing worker can’t be certain that they’ve fully inspected a small screw or diesel injector nozzle if the human eye isn’t capable of detecting the slightest variations in thread spacing. And a camera alone can’t make that call. AI analysis of an image is needed before a decision can be rendered about whether to clear for shipping or pull out of inventory. That’s why companies like Bosch have come to Zebra for help designing and onboarding an AI-powered solution that can effectively pack three inspection and verification processes into a single Zebra-Matrox vision system.
But AI isn’t just helpful from a quality control perspective. It’s critical to inventory management, too. A manufacturer can’t know with any degree of certainty what SKUs to prioritize or the quantities they need to produce in a certain period if AI isn’t tracking different market dynamics and telling them what to produce and when. That’s why stock levels have been so imbalanced with demand for so long. Most CPG companies have been slow to embrace AI even though the use case and value proposition has been well proven with large brands such as Grupo Bimbo and Estee Lauder.
Closer to home, the consequences of not using AI could be most evident in the front-line workforce. A grocery store worker may not know if they should be stocking shelves, shopping curbside pickup orders or staffing a register if not for AI analyzing store operations along shopper demands and prioritizing tasks for that worker. Nor will they know if an item is out of stock on a shelf for hours on end unless the camera pointing at the shelf is feeding into an adaptive AI that can extract key learnings from that footage, such as missing items, and alert the right person to address. In fact, we have customers using AI to detect in-store inventory and sales anomalies via Zebra’s Prescriptive Analytics solution.
So, the next time someone makes it sound as if AI can be harmful, reach out to us at Zebra to talk about all the ways it could possibly be helpful. Find out how it can help you and your workers see more, do more, make better decisions and take faster action. Better yet, ask if there is AI-powered software available to help augment or fully automate all decisions that must be made in support of your front-line operations. Because at the end of the day, the whole point of AI is to make us better at our jobs, not take our jobs.
That’s why our AI researchers are on a mission to ensure Zebra and its customers are prepared for the rapidly growing desire and need to deploy and run AI solutions at the edge. It’s also why more attention needs to be put on how AI is helping people by helping businesses.
The growing number of AI applications we’re working on at Zebra are all fueled by the challenges of doing business in every industry today. Our teams work on various projects, including object/product detection/recognition, retail shelf analytics, robotics, analyzing shelf health, orchestrating agents in retail and warehouse spaces, digital twins of physical spaces, edge deployment and voice assistants. And our researchers are pioneering ways to infuse AI into more of our hardware and software portfolio to benefit front-line workers in every industry, from warehousing and manufacturing to retail and healthcare.
Our vision is to bring different AI applications (voice, vision, text, sound) together (multimodal AI) in the same way a human uses their five senses, to sense and analyze the world around them to inform decision-making and action (act). Without AI, it’s going to be hard to make everyone – or anyone – happy given our recognized human limitations and the tremendous demands people are placing on one another.
Plus, as the number of smart devices – IoT physical objects – at the edge of networks continues to grow, we know responsible and ethical AI is going to become as necessary to business as the smartphone is to your personal livelihood. AI is going to be called upon more and more to help us make sure all the data your business is collecting is worth something.
That said, we know the concerns around generative AI extend to all AI, and that we must take care to consider the implications of new AI use cases that are brought forth. That’s why every decision we make related to AI use within Zebra’s solution offerings is also driven by critical factors such as data privacy needs, the latency of responses, cost to compute and network bandwidth availability. It’s also why we continuously evaluate and refine our responsible AI methodology in terms of ethics, development and deployment and will evolve processes, principles, tools and training while ensuring consistency and compliance through an internal hub-and-spoke governance model – much like we do for cybersecurity. We are committed to responsible, transparent AI use.
In fact, we have just compiled our first set of AI ethics principles, and we support the Business Roundtable’s (BRT) AI Road Map and policy recommendations for responsible, appropriate, and ethical AI development and deployment. Watch this: