In Philip K. Dick’s short story “Autofac,” which appeared in the November 1955 edition of Galaxy Science Fiction, tungsten shortages in Pittsburgh and Detroit pinch the network operating autonomous underground factories around the planet.
The Autofac network searches far and wide for scrap metal to fill self-driving ore carts. Governing the whole system, artificial intelligence remains oblivious to humans. “We’re licked,” one character says, “like always. We humans lose every time.” The primary concern is the factories’ self-replication, no matter the cost to the environment or people.
The story appeared a year before AI research got started with a conference at Dartmouth College. Attendees proposed to “make machines use language, form, abstractions, and concepts to solve all kinds of problems now reserved for humans, and improve themselves.” The name “artificial intelligence” was coined at this time.
More than 60 years and a couple of developmental pauses later, AI is infused in every aspect of manufacturing, from product design to distribution. And Dick’s sci-fi nightmare looks archaic.
“In many cases, AI is freeing up time, creativity, and human capital, essentially letting people work more like humans and less like robots,” write Paul R. Daugherty and H. James Wilson in “Human + Machine: Reimagining Work in the Age of AI.”
“By collaborating with robots, we will reach new levels of productivity with more adaptive assembly lines. The safer, less bored worker will update a co-bot on each shift, maybe entering some new code. If things get sticky, a data hygienist — one of the new specialized jobs to be created — is on hand in the plant. Even back office functions — order processing and production runs — will benefit from AI gnawing through large amounts of data and, as a big bonus, making decisions.”
A “lights-out” factory operating on its own with just occasional checks by humans might still be a little far out, although FANUC, which has its American headquarters in Rochester Hills, has had robots building robots since (fittingly) 2001. Even as we embrace the current level of manufacturing technology, Industry 4.0, obstacles remain before there can be full acceptance and implementation of important tools.
“The biggest challenges that we’re seeing are in good data-collection,” says Chuck Werner, lean program manager at the Michigan Manufacturing Technology Center in Plymouth Township. Imparting Six Sigma practice fires Werner up. “Most places, if they’re collecting data, it’s manual. Some are still not collecting data.”
Industry 4.0 requires a big push among manufacturers to become real-time, data-driven organizations, Werner says. The first step is to identify the best data to collect, which means broadly installing sensors before looking at key performance indicators like uptime, throughput, and costs.
“Large manufacturers can throw sensors on everything, try to figure out which are important, then reproduce only the things they need to be collecting,” Werner says. The problem is, many small- to medium-sized manufacturers lack the R&D, and maybe even the initiative, to sprinkle sensors around like salt.
Another bugbear is in implementing technology already at hand. Werner has seen companies buy programs to automate things such as purchasing raw materials, scheduling workflow, and tracking inventory — yet a paucity of training on these programs can lead to limited use and little return on investment.
He summons up a manager’s voice: “I can’t see the reports I want to see, and nobody knows how to modify the program to get them. So I’m just not going to use the system — I guess I’ll go back to Excel.”
As an example of a manufacturer “well down the lean journey,” Werner cites Brembo’s 750,000-square-foot operation in Homer, southwest of Jackson. Brembo, which was formed in Italy in 1961, makes brakes for the seventh-generation Chevrolet Corvette and other high-performance production vehicles. Acquiring the former Hayes Lemmerz International operation in 2007, Brembo has expanded at the site, including a $100-million investment to open its first North American cast-iron foundry in 2016. The company’s North American headquarters in Plymouth Township houses R&D, sales and marketing, and central staff units. Altogether in the United States, Brembo employs about 800 people.
“I think the plant has done an excellent job working on several of the advanced manufacturing principles,” says Dan Sandberg, president and CEO of Brembo North America, who has had responsibilities at the Homer plant since 1999 and now oversees operations in Mexico, as well. “They use a lot of technical tools to improve the flow and flexibility of the plant. It’s very clean. The automated process flow is very well-drawn-out, so you don’t have a lot of wasted movement.”
Brake rotors are heavy, of course, so Brembo has robots to do the lifting. Raw materials go in one door and precisely finished products get loaded onto trucks outside another. “We sensor almost the entire line operation. A single machine may give us 15 to 20 items that we’re tracking,” Sandberg says.
Data accumulates and evaluations are made on such factors as vibration, temperature, weight, and cutting dimensions. What’s more, AI removes human judgment about adjustments and replacements from the equation. “Based on data that we give them that shows the variation in our machining process, the machine will automatically adjust to the pressure of the tool when it sees that the process is deviating,” he says. “You don’t want people making a change in the tooling process just based on how they’re feeling that day.”
It shouldn’t be taken personally. In fact, with Industry 4.0, workers have a new array of implements and tasks, along with added authority. A tablet computer is essential for many station operators, who collaborate with AI robots in customized production. “That robot is working side by side with an employee in doing some of the things that a person may be doing,” Sandberg says. “We don’t like our employees to be loaders and unloaders. That’s terrible work. It’s work that’s boring and not very challenging.” Instead, the operator is managing robots and doing maintenance on the systems.
Predictive maintenance is especially crucial for Brembo because, as Sandberg says, “All of our capacity is pretty precious for us.” At the foundry in particular, unanticipated downtime causes a chain reaction — or, as he calls it, an I Love Lucy situation — throughout the plant.
Brembo is a large company that operates on four continents and has a workforce of 9,000 people. For small- to medium-sized companies that can’t afford high-priced programmable logic controllers to interface with production machines, startups like Oden Technologies in New York City provide new ways to analyze real-time efficiencies. Companies also face obstacles in adding sensors everywhere, to network through the Internet of Things.
A 2018 report by CB Insights looked at this question and concluded, “Slapping cheap IoT sensors on everything isn’t a cure-all, and it’s entirely possible that more value gets created from a smaller number of more specialized, highly accurate IoT sensors.” To answer this question, Augury, an Israeli startup, provides machine-health solutions to more easily perform predictive maintenance.
With an increasing emphasis on modular production methods and more customized goods — Brembo offers its brake calipers in a plethora of colors — CB Insights foresees a day when “humans are largely uninvolved with manufacturing.” The report asserts, “Presumably, tomorrow’s manufacturing process will eventually look like one huge, self-sustaining cyber-physical organism that only intermittently requires human intervention.”
That means the “Autofac” of Dick’s imagination is coming true. But is the lights-out
factory and elimination of most related jobs really that close?
“It doesn’t bear out in the data,” says Tom Kelly, executive director and CEO of Automation Alley in Troy, a nonprofit think tank that bills itself as Michigan’s “Industry 4.0 Knowledge Center.” Kelly says a fall-off of jobs is worth watching for, “but every time robot sales go up, jobs go up. Every time robots sales dip, jobs dip. Companies apply robots to become more efficient because they see, on the horizon, more orders coming in. So as they become more efficient, they’re able to take more orders. Even though they’re using less people per order, they need more people.”
The greater short-term problem may be that executives and managers of small- and mid-sized companies are resisting change. It could be due to apathy or limited technical background, but failing to move forward on automation means “we’re going to stagnate and decline.”
Kelly recommends two things to prepare for the accelerating trend of AI integration. “If you’re in a position of leadership, you really need to understand what’s happening in the technology space, and then start to make educated guesses on what could disrupt your business as you know it,” he says.
Depending on the chief information officer may not be enough. AI, 3-D printing, sensors, and big data could be disruptive forces, and the organization’s leader must ask how to ameliorate the problem. Outside help is available. “Don’t be the frog that boils in the water,” Kelly says. “Develop a plan for how you will act, and how you will mitigate the risks.”
Part of that plan could involve managing changes that eliminate the most menial tasks. A 2017 New Yorker article, “Welcoming Our New Robot Overlords,” introduces the topic. The magazine visited Steelcase, the large office furniture manufacturer in Grand Rapids, and workers there said automation had made the factory “cleaner, less noisy, and more productive.” They were less bored and less physically taxed, as well.
“So the question becomes, why continue to train people to work like robots?” ask the co-authors of “Human + Machine.” Their rosy picture of co-bots on the production line includes postulations about new positions that will be created for people. The Steelcase workers spoke of on-the-fly fixes in their plant.
“When something went wrong with the assembly (process), they could diagnose the problem swiftly by consulting the data,” the article says. This observation anticipates one of the jobs for advanced AI — an algorithm forensics analyst. The analyst, say Daugherty and Wilson, will conduct an “autopsy” to explain a system’s logical mistake.
A company of the right size should also have a transparency analyst who classifies why an algorithm acts as a “black box” — that is, protects intellectual property or is simply a critical piece of code and therefore hard to access. The transparency analyst’s database will be invaluable to the explainability strategist, who determines which AI technology is best for the application.
“In many industries and countries, the most in-demand occupations or specialties did not exist 10 or even five years ago, and the pace of change is set to accelerate,” according to the most recent World Economic Forum report on the future of jobs. “By one popular estimate, 65 percent of children entering primary school today will ultimately end up working in completely new job types that don’t yet exist.”
Thinking of jobs that have been displaced by technology in combination with social shifts, consider the gas lamplighters of the 19th century, who were displaced by electric lights. Other examples include railroad brakemen, newsroom copy boys and print shop typesetters, gas station attendants, bowling alley pinsetters, and baseball park scoreboard operators — not to mention all the manual assemblers who have disappeared in the mists of time.
But in a conundrum that experts are still trying to solve, staffing today is the biggest problem for many companies. Consider Michigan’s manufacturing sector is one of the most advanced in the world, and yet the state’s unemployment rate is 4.1 percent (October 2019).
“The simple truth is that companies can achieve the largest boost in performance when humans and machines work together as allies, not adversaries, in order to take advantage of each other’s complementary strengths,” write Daugherty and Wilson.
Rather than being the threat envisioned by Philip K. Dick, AI holds the promise of expanding productivity — and the wealth that goes with it — as well as pushing humans to achieve their fuller potential.
Jobs Of The Future
Trainers
Computers learn to adapt to humans. People teach them things like empathy and world views. Some computers might not understand, though. Screaming at the phone about the performance of the Cincinnati Bengals football team still gets you to their billing department.
Explainers
Bots coach managers, too. Someone has to explain to Sleepy Sam, head of purchasing, how the water will continue to sweep over his bow unless he starts using internal combustion engines instead of oarsmen. And stop blaming human resources for sending over extra people.
Sustainers
Meet the AI safety engineers, who offer a cheery, machine-like refrain to any human question. They might respond: “We’re here to allay distrust. Your inquiry is not a problem. Awesome! I’m reaching out to our machine relations manager right now and we will solve your problem lickety-split.”
Tune In
Automation Alley, a nonprofit think tank in Troy that bills itself as the Industry 4.0 Knowledge Center for Michigan, centers its focus on future productivity offerings. “We’re trying to look around the bend and over the hill about what technology is coming,” says Tom Kelly, executive director and CEO of Automation Alley.
Yes, Industry 4.0 will affect manufacturing, but also the health care, agriculture, and retail sectors. DBusiness recently spoke with Kelly about what’s happening.
DB: Who are smart manufacturing leaders?
TK: Obviously, looking at the OEMs, Ford, GM, and FCA have been leaders for decades. They don’t get the credit they deserve. “Automation” and “smart factory” have been synonymous with the Detroit area for a long time. The (Big Three) have done a very good job of driving out waste, automating as much as they could, and finding that balance between automation and human labor. If you go down the food chain, you see companies like Hirotec America, a door and closure manufacturer in Auburn Hills. They sit on our board of directors. They’ve done a wonderful job in smart manufacturing. Most of the people in the region don’t do smart manufacturing very well because they don’t really understand it.
DB: Do you see small and medium companies without in-house R&D outsourcing AI development?
TK: The jury’s still out on what’s going to happen to the small and middle markets. As technology rapidly expands, one of the risks is that the benefits of technology and AI accrue to the very wealthy, i.e., the biggest corporations that can afford them. We’ve seen that happen in the purely digital side. In Silicon Valley, the first movers tend to be monopolistic: There’s one search engine, Google; there’s one social network, Facebook; there’s one online movie platform, Netflix; there’s one retailer, Amazon. So it’s not theoretical that all the wealth could accrue to a very few. Our job at Automation Alley is to make sure that doesn’t happen, especially in Michigan. We’re trying to protect that (supply and distribution) chain. Small businesses, by their very nature of being more innovative and agile, have the ability to survive and play a role going forward.
DB: What frustrates you?
TK: What’s frustrating is the apathy and lack of desire to understand and want to participate in the future. We’re having a dickens of a time getting the smaller players to pay attention. We’re shouting “Fire!” in the theater. They’re telling us, “Get lost, we want to finish the movie.” That’s our pain, and our shame. We can’t get them to move faster.
