“I think the ratio of humanoid robots to humans will exceed 1:1,” Elon Musk said on March 1st. This sentence, spoken from Tesla, the self-proclaimed king of electronic sound, is more like a promise than a prediction. Mr. Musk’s car company is developing an AI (Artificial Intelligence) robot codenamed Optimus, which will be used in homes and factories. When he said this on Tesla Investor Day, there was a video of Optimus walking around without assistance behind his back.https://store.stoneitech.com/
Given that Mr. Musk did not elaborate on how or when the promotional film would transform into an army of over 8 billion robots, we can consider this as science fiction for the time being. But he did get involved in very real discussions about the future of the job. The automation of specific forms of AI empowerment is rapidly transforming into scientific reality.
Since November last year, AI interlocutor ChatGPT has stunned users worldwide with its ability to almost impersonate humans. By analyzing a large amount of data on the Internet, other “generative” AI’s are also creating text, images, and sounds similar to human works. Last month, the boss of computing giant IBM predicted that AI would be able to complete the work of most white-collar employees. On March 6th, Microsoft announced the release of a series of AI “assistants” that can assist employees engaged in various tasks such as sales, marketing, and supply chain management. The restless observers are already muttering that the world is about to end at work.
Of course, there has been a long-standing concern about the job substitution effect of technology. In Britain in the early 19th century, Luddites who strongly resisted technological innovation burned down factory machinery. With the application of wartime mechanized innovation, which sparked a wave of panic about mass unemployment in the 1950s, the term “automation” became widely known for the first time. In 1978, British Prime Minister James Callaghan requested the government to conduct research on the potential for job losses in the technological breakthrough of his era – microprocessors. Ten years ago, Carl Frey and Michael Osborne of Oxford University published a famous article that was subsequently cited 5000 times, claiming that in the next “ten or twenty years”, 47% of the tasks performed by American workers will be automated. Now, even the optimistic Mr. Musk is wondering what it means to have more robots than humans: “Even then, it’s not clear what the economy itself is.”.
It will take several years to determine whether Mr. Frey and Mr. Osborne are correct, and Mr. Musk’s views can also be completely ignored. Early fears of technology harming employment have always been unfounded. On the contrary, the labor market in developed countries has historically been tense and has become structurally tightened as societies age. “For every unemployed American, there are now two jobs to fill, a record high in history.”. The manufacturing and leisure and entertainment industries in the United States have labor shortages of 500000 and 800000 people, respectively.
Therefore, the most urgent issue for developed economies is not too much automation, but too little. To make matters worse, for large companies, the implementation of automation is generally more difficult, and it is likely to be no simpler than the recent boom in AI.
In factories, robotic arms that perform repetitive tasks such as welding, drilling, or moving objects have existed for decades. Historically, robots have been used intensively in the automotive industry because heavy components and limited types of large batches of work are ideal for machines. The electronics industry needs precise and repetitive actions, and it is also an industry that has earlier adopted machines.
According to Jeff Burnstein, president of the American Association for the Advancement of Automation, the list of industries embracing robots has recently grown longer. Sami Atiya, who runs the robotics business at ABB, a Swiss industrial company, points out that advances in computer vision have made machines more flexible. Lightweight “collaborative robots” are no longer caged, but work alongside humans. Automated vehicles can also transport items from one place in factories and warehouses to another.
At the same time, the price of robots has also fallen sharply. Asset manager Ark Invest estimates that the average price of industry robots decreased from $69000 in 2005 to $27000 in 2017. Last December, ABB opened a “super factory” in Shanghai to let robots build robots. Susanne Bieller, Secretary General of the International Federation of Robotics (IFR), pointed out that the new “code free” system does not require programming skills, resulting in lower installation costs.
Thanks to technological progress and falling prices, the global population of industrial robots has increased from 1 million in 2011 to nearly 3.5 million in 2021. Sales of Fanuc, a large Japanese robotics manufacturer, rose 17% year-on-year last quarter, while Keyence, a Japanese company that provides automation consulting for global factories, also increased 24%. Although its sales volume has declined compared with that of 2021 when there was a foam (at that time, CEOs were looking for alternatives for the shortage of manpower caused by COVID-19), the share price of robot manufacturers is still one fifth higher than that before the epidemic.
Despite some growth, the level of adoption of robots is still low, especially in Western countries. According to IFR analysis, even South Korean companies that have adopted robots around the world (one robot for every 10 manufacturing workers) are a long way from Mr. Musk’s vision. In the United States, China, Europe, and Japan, the ratio is 25-40 to 1. According to Boston Consulting, the world spent $25 billion on industrial robots in 2020, which is less than 1% of global capital expenditure (excluding the energy and mining industries), and even less than the money people spend on sex toys.
Rainer Brehm, head of factory automation at Siemens, a German industrial giant, believes that the longer life cycle of industrial equipment limits the speed at which older, dumber machines can be replaced by smarter new ones.
Currently, in developed economies, the least skilled jobs are still in services where human tasks are difficult to automate. The human body is a flexible and flexible biological miracle that can achieve 244 planar movements through joints and fingers. Kim Povlsen, president of Universal Robots, an industrial robot arm manufacturer, pointed out that the average robot has only six such “degrees of freedom”.
Due to similar old systems and corporate inertia, the automation of office work is also progressing slowly. In theory, digitization should free people from daily tasks such as ordering inventory, paying suppliers, or aggregating accounts.
Cathy Tornbohm of Gartner, a research firm, believes that in fact, many pre digital era companies use intertwined outdated and incompatible systems. Many companies prefer to outsource these chores to low cost countries such as India or the Philippines rather than hire IT consultants to sort out the mess. According to another research company, IDC, the software cost for solving cumbersome office tasks in the entire market is only $20 billion per year, which is less than the money spent on purchasing physical robots.
Over time, further innovation may remove some of these obstacles. From the perspective of physical robots, this is happening in South Korea, which is obsessed with machines. Doosan Robotic, one of the country’s largest robot manufacturers, has opened its software to external developers to create pre programmed applications for its robots, ranging from making coffee to laying floor tiles on construction sites. Robert Chicken uses a robotic arm to operate its frying pan. To help reduce its initial investment in its stores, Doosan Robotic rents robots to it at a price of approximately $900 per month, which is significantly lower than the labor cost of human operators. South Korean internet giant Naver has a dedicated department that develops robotic vehicles that can shuttle through busy environments with complex layouts, and a large army of such robots has already sped past delivering lunch boxes and packages to its employees.
Office automation is also becoming more advanced. UiPATH is a pioneer in automating simple tasks such as copying and pasting between different programs. The company is now able to provide tools such as using image recognition algorithms to extract data from written documents, and drawing business processes by observing employees’ operations on computers. Rob Enslin, co president of Uiperth, said the company has served 10000 customers. Charles Lamanna, the head of automation products at software giant Microsoft, pointed out that the company’s Power Automation tool now has 7 million monthly active users, and can help ordinary white-collar workers with little programming experience achieve automation such as expense or travel review.
Some companies are also beginning to tentatively embrace generative AI. However, just like robots and process automation, the implementation of new technologies will not be achieved overnight. Allen&Overy Law Firm released a virtual legal assistant with ChatGPT like functionality in February, requiring lawyers to cross verify all responses from robots. The technology news website CNET secretly released 73 articles written by a robot last November. The reporters were initially dismayed, but later they felt very funny because these articles were found to be prone to errors.
Mr. Lamanna believes that AI technology behind chat robots can help automate in the future. However, moving from science fiction to scientific reality is one thing, and moving from scientific reality to economic reality is quite another matter.