This is the next article in our ongoing series, Eyes on Industrial AI, where we explore the multifaceted world of industrial artificial intelligence and automation. Let’s uncover the innovative applications of these cutting-edge technologies specifically as they relate to manufacturing and supply chain management in today’s era of Industry 4.0, including any challenges they encounter.
To the average consumer, it seems that robots have been out to “steal our jobs” ever since grocery stores invested in self-checkout stations. Many raised alarms about the cashiers and checkout clerks who would be replaced by automatic scanning machines and sensors.
Since the introduction of self-checkout technology approximately ten years ago to today, the grocery store still isn’t fully manned by robots. Why aren’t robots taking more jobs from humans?
In reality, most grocery stores use self-checkout as an additional organizational process. For example, self-checkout lines will be opened to help deal with the busy customer rush around certain hours.
You’ve probably also noticed that grocery stores use a combination of self-checkout and traditional checkouts to give employees more freedom to address other necessary tasks, like training, restocking, or cleaning.
Also, even though the technology has progressed greatly over the few years it’s been popular, self-checkouts still need at least one human employee supervising the operation. This supervisory role is a newly created position only through the adoption of cutting-edge technology.
So what do self-checkout machines have to do with automation trends in broadscale manufacturing?
Think of a self-checkout station as a factory microcosm. We’ll go into several reasons below why we believe that AI technology will definitely eliminate some traditional jobs, but the supervision and mastery of operating high-tech machine systems will create more jobs that haven’t yet even been theorized.
In short, are robots really going to steal human manufacturing jobs? Not according to the evidence, so let’s discuss how and why we’ve come to this conclusion.
AI has been a hot topic, and doubts over automated technology represent very big gaps in understanding practical AI applications. Things like self-driving cars and the recent question of LaMDA’s sentience are very sci-fi, and don’t represent the reality of robots in areas like industrial manufacturing.
There’s definitely a wave of top manufacturers moving towards automated processes, and it’s moving quite quickly. According to Stanford’s review of the 2022 AI Index, the major takeaways about AI trends are increased number of patents filed, lower costs of AI usage, and a major spike in interest for AI ethics research. Overall, the Index reports that private investment in the field of AI “more than doubled” since 2020. Unlike more spurious trends like crypto and NFTs, AI tech actually delivers, with quantifiable improvements in quality control and process improvement.
The World Economic Forum extrapolates, estimating that “85 million jobs will be displaced while 97 million new jobs will be created across 26 countries” by 2025.
“The impact and benefits of AI will likely not be shared equally. Businesses and governments must work together to ensure that as many people as possible can benefit and the digital divide does not increase and exacerbate existing inequalities.” - World Economic Forum.
AI technology is already at the forefront of industry, it’s just a question of how it is implemented.
Automation in manufacturing is hugely driven by labor shortages in the modern economy. Particularly during the pandemic (including long-lasting effects of illness), large numbers of workers were unable to perform their jobs due to illness.
As workers fell sick, and others were laid off or lost their means of employment, there was a great loss of skill development over time within companies. Turnover increased, and workers left or changed industries, leading to more day-to-day uncertainty in workflow.
Even if the “robots” used to fix this issue are integrated software elements adjusting to absenteeism in real-time, they expertly address this new complication.
Whereas the self-checkout machines mentioned in the intro are generally described as janky or difficult, so much money and research has gone into developing highly specialized tools for quality assembly that they are infinitely more usable. It’s almost as if there’s two whole different worlds of artificial intelligence!
Google’s already developed a Natural Language Processing (NLP) system, but after only a few years, they’ve introduced an exponentially more powerful system called BERT, and another called MUM, that can process syntax and language peculiarities. This goes to show that certain niche types of AI tech attract a supercharged development path due to their potential in industrial revolution.
As technology improves and research requirements get more precise for higher quality control, there is a greater need for higher-precision operations. A screw that functioned adequately in a small ship, for example, may not be up to par when placed as a key component of an extensive oceanographic rig, for example. That kind of quality precision is often only achievable through industrial equipment.
Moreover, sometimes mechanical movements can be so difficult that not just anyone can operate the machinery – specialized welders and other tradespeople have the practical experience to judge the correct application of certain tools.
When we think of AI technology and robots together, we might picture something that looks kind of like The Terminator or the Boston Robotics dogs. But what do actual AI “robots” look like in manufacturing situations?
Of course there are machines that are pretty obviously robotic, like articulated arms that independently solder and perform highly detailed engineering tasks.
However, robots can exist in many other forms just by nature of relying upon automation and artificial intelligence:
From a more realistic perspective, automation and artificial intelligence are technologies that are more present in basic tasks than in the few fanciful cases people hear about on the cutting edge.
The Wall Street Journal argues that AI hype has overblown actual capabilities: “AI ethicists and researchers warn that some businesses are exaggerating the capabilities -- hype that they say is brewing widespread misunderstanding and distorting policy makers' views of the power and fallibility of such technology… In reality, artificial intelligence encompasses a range of techniques that largely remain useful for a range of uncinematic back-office logistics like processing data from users to better target them with ads, content and product recommendations.”
Put simply, if you were hoping that your new robot coworkers would be more like C-3PO than a bland database, you’d be sorely mistaken.
Okay, now that we’ve set the foundation by explaining why there’s such a growing trend of automation technology in manufacturing, let’s examine some areas of concern and match them to actual data insights.
A Forbes Technology Council made of senior-level tech executives went on record to name the most likely jobs that will be replaced by automation: insurance underwriting, warehouse and manufacturing jobs, customer service, research and data entry, long haul trucking, and “Any Tasks That Can Be Learned.”
Now, this obviously sounds like a death sentence; “customer service” and “research” categories alone make up an insurmountable number of jobs alone! However, "at risk" doesn't mean complete elimination of a subset of the workforce:
Back in the DOS days, the Forbes Council notes, “All it took was one hyper-dedicated ‘rogue’ and six or so months of toil to create a game that could be disseminated as shareware. Now it takes a high-tech village to produce what characterizes a ‘movie studio-like undertaking’ requiring voice talent, designer, sophisticated physics and a multi-million-dollar budget.” Forbes Technology Council.
The Guardian reports that labor shortages in part are driving the switch to AI systems, which further pushes out white-collar workers. Notably, the types of jobs that still remain safe from AI upheaval include three types:
Interestingly, “knowledge work will prove to be easier and less expensive to automate than lower-paid work that requires physical manipulation,” the report states. This means that we may think of the “lowliest” menial jobs like cleaning or restocking as ones that can be automated, but it’s more likely that robots will replace the data entry clerk before the janitor, given the discrepancy between routines and unpredictable tasks.
This brings us to the second, lesser known type of job automation will target: superfluous management roles.
A published and well-reviewed study on AI development titled, “The Robot Revolution”, found some particularly insightful results: Investments in robotics are…
The findings of this study imply that the greatest subsector of the manufacturing workforce at risk to automation is overlapping or unnecessary middle management. Backing up this claim is the lean manufacturing philosophy of reducing bureaucracy as a form of waste. If this remains widely true, then low-skilled workers will be aided by automation, seeing as they will gain more supervisory control and clearer paths to leadership positions.
A key factor of the data is the geographical distribution of manufacturing jobs. Some cities’ main industries are in manufacturing, and the majority of the population works at the same production sites. In these places, a sudden jump to automation (especially without reskilling workers) could be disastrous for both the local economy and the factory in question.
The discrepancy between local and national data is shown to be significant in MIT’s study on the impact of robots on the job market (emphasis added):
“Between 1990 and 2007, the increase in robots (about one per thousand workers) reduced the average employment-to-population ratio in a zone by 0.39 percentage points, and average wages by 0.77%, compared to commuting zones with no exposure to robots, they found. This implies that adding one robot to an area reduces employment in that area by about six workers. “But what happens in one geographic area affects the economy as a whole, and robots in one area can create positive spillovers. These benefits for the rest of the economy include reducing the prices of goods and creating shared capital income gains. Including this spillover, one robot per thousand workers has slightly less of an impact on the population as a whole, leading to an overall 0.2 percentage point reduction in the employment-to-population ratio, and reducing wages by 0.42%. Thus, adding one robot reduces employment nationwide by 3.3 workers.” - “Robots and Jobs: Evidence from U.S. Labor Markets”.
The quote above is a perfect example of the need for specificity in strategic operations. When assessing your working AI capabilities and the effect on the company, you need local and industry-specific data to guide your implementation process.
Even if adding automation or AI to your assembly operations is an obvious next step to process improvement, communicating the extent and necessity of human teamwork will help set up strong foundations for the future.
So throughout this article we’ve challenged some of the latent assumptions about the nature of AI in the manufacturing workforce, and found the following conclusions:
According to this Wharton Study about robots in human-led roles, scientists reported, “Any employment loss in our data we found came from the non-adopting firms. These firms became less productive, relative to the adopters. They lost their competitive advantage and, as a result, they had to lay off workers.”
It’s clear that not only is there a soft advantage in pursuing AI due to marketing and investment opportunities, but there is also a hard data advantage in pursuing AI according to the quantitative results of this study. You may not love the new world of artificial intelligence in manufacturing, but it is already here, now, and ignoring it would be a detriment to your business’ success.
The World Economic Forum (linked above) maintains that AI will automate repetitive and dangerous tasks, and names data entry and assembly line manufacturing as examples. These roles, however, are far too broad from which to draw any actionable conclusions.
After all, what kind of assembly line manufacturing are we even talking about?
Most of these reports, while reputable, go with averages and mitigatory statements rather than industry-specific know-how. In addition, just because robotics are becoming more universally applicable doesn’t mean that human soft skills are becoming less valuable.
Chances are you’re going to need experienced trades workers to understand exactly how to handle new automated equipment. For example, at the most recent Fabtech conference, AGT introduced the LayoutMaster, a new machine technology that saves welders hours in cycle time. Even though it can be operated by someone without welding experience, the machine would still “benefit from someone with welding skills to operate.” An AGT engineer explains, “You’re still dealing with a piece of equipment where something could go wrong. You need to know whether it’s doing a good job. As a welder, you know what to listen for.”
“Defective welds can cost metal fabricators a lot of time and resources to rework, so skilled welders can prevent the large-scale production of incorrect welds through early detection of flawed techniques.” - Leland Weed, AGT Robotics.
This is where creative solutions involving robotic potential and human design come into play. Things like work instruction software for manufacturing can amplify the abilities of skilled trades workers, allowing them to have safer, faster, higher quality production cycles while still preserving the human touch.
AI is becoming more specialized, but so far only has the capacity to aid and streamline human jobs in manufacturing, rather than replace broad swaths of the workforce. By adopting lean long-term strategies for protecting your workers and your market share, you can deftly adopt automation without fears of robots eliminating human working potential.