For decades we believed skilled trades and physical labor would be the last frontier of automation. Recent advances in physical AI suggest we may need to rethink that assumption—not because robots are becoming human, but because they may not need to.
From the Craig Bushon Show Media Team
For generations, people have taken comfort in the idea that skilled trades and physically demanding occupations would remain among the safest careers in an increasingly automated economy. Manufacturing jobs might become more robotic. Warehouses might rely on more machines. Office work might be transformed by artificial intelligence. But plumbers, electricians, mechanics, carpenters, and countless other tradespeople seemed protected because their work depends on something machines have historically struggled to master: moving through unpredictable environments while manipulating the physical world with precision and judgment.
That belief was reasonable based on the technology available at the time. Today, it deserves a second look.
One phrase heard on almost every job site provides an interesting starting point. A mechanic working under a vehicle says, “I wish I had another set of hands.” An electrician asks an apprentice to hold a panel while terminating wires. A plumber needs someone to steady a section of pipe while making a connection. A carpenter asks a coworker to support the far end of a beam during installation. These moments are so common that we rarely think about them, yet they reveal something important about the way physical work is performed.
Those everyday frustrations led us to what we are calling The Third Hand Theory.
The Third Hand Theory begins with a simple observation. We have been evaluating robots by asking whether they can replace a human being. That comparison assumes the goal of robotics is to duplicate human anatomy and human capability as closely as possible. Engineers are not bound by that assumption. Human beings have two arms because God produced that design. A robot has whatever number of arms, sensors, cameras, or tools engineers decide will maximize productivity for a particular task.
Imagine a service robot repairing plumbing inside a commercial building. One arm could stabilize the pipe without moving. Another could illuminate the workspace with perfect precision. Two additional arms could manipulate tools, while integrated sensors continuously monitor alignment, torque, and pressure. The objective would not be to imitate a plumber as faithfully as possible. The objective would be to turn a two-person job into a one-person job—and to finish it in an afternoon instead of a day. Suddenly, the familiar expression, “I wish I had a third hand,” no longer sounds like a figure of speech. It begins to sound like a design requirement.
Thinking about robotics this way changes the conversation. Instead of asking whether a robot can fully replace a licensed electrician or an experienced mechanic, we begin asking whether robotic assistants can perform enough supporting tasks to make one skilled worker substantially more productive. Can they carry materials, position equipment, perform inspections, retrieve tools, or hold components in place while the human focuses on the work that requires experience, licensing, and judgment? If the answer becomes yes, the economics of labor begin changing long before a robot is capable of replacing an entire profession.
This distinction matters because businesses rarely evaluate technology in all-or-nothing terms. They evaluate productivity. A company does not need a robot that performs every aspect of an electrician’s job. It needs technology that reduces labor hours, shortens project timelines, improves consistency, or allows one experienced employee to accomplish significantly more during a workday. History shows that disruptive technologies often spread because they improve part of a workflow before they eventually reshape the entire process.
The deployment record is beginning to bear this out. At BMW’s assembly plant in Spartanburg, South Carolina, two humanoid robots built by Figure AI worked an eleven-month rotation on an active production line—ten-hour shifts, five days a week—loading more than 90,000 sheet metal components and contributing to the production of over 30,000 vehicles, according to company disclosures. In Georgia, Agility Robotics’ Digit robots have moved more than 100,000 totes at a GXO-operated warehouse, part of a fleet that has now logged over 65,000 operating hours across nine facilities per the company’s securities filings. Mercedes-Benz is testing Apptronik’s Apollo robot in automotive manufacturing. Notice what none of these robots is doing: replacing a skilled tradesperson. Every one of them is doing precisely the kind of supporting work the Third Hand Theory describes—moving materials, loading parts, filling the repetitive gaps between human workers and existing automation. The theory is not a forecast of some distant future. It is a description of the business model already generating revenue.
Another factor receiving less public attention is the way knowledge itself may be distributed. Experienced tradespeople accumulate valuable expertise over decades, traditionally passing it to apprentices through observation and repetition. Today’s leading robotics companies are building their systems on imitation learning—training machines by recording skilled human workers performing tasks, then converting that recorded expertise into software. Once a technique is successfully learned and validated, it can be replicated across an entire fleet of machines through an update, the way a phone receives new features overnight. An apprenticeship transfers knowledge to one person over several years. A software deployment transfers it to ten thousand machines in an evening. That asymmetry would not eliminate the need for experienced human workers, particularly in complex or unpredictable situations, but it raises a question worth sitting with: whose accumulated skill is training these systems, and who captures the value when that skill becomes software?
Viewed through this broader lens, the discussion extends far beyond robotics into economics, education, industrial policy, and national competitiveness. Here the numbers deserve close attention. Chinese firms accounted for more than 80 percent of global humanoid robot installations in 2025, according to Counterpoint Research. A single Chinese manufacturer, Unitree, shipped roughly 5,500 humanoid robots last year—more than every American competitor combined—while driving its entry-level pricing below $16,000. Average humanoid prices industry-wide have fallen from about $85,000 in 2023 to roughly $25,000 in 2025. Countries investing aggressively in artificial intelligence, advanced manufacturing, robotics, and energy infrastructure are not pursuing separate objectives. They are building complementary systems designed to reinforce one another over the coming decades.
The United States continues to lead in many areas of scientific research, entrepreneurship, software development, and technological innovation, and American firms arguably still hold the edge in the AI systems that make these machines useful. History reminds us, however, that inventing a technology and deploying it across an economy are not always the same achievement. The country that invented the lithium-ion battery does not dominate battery manufacturing today. Commercial adoption, manufacturing capacity, workforce preparation, infrastructure, and public policy all influence who ultimately benefits most from technological change—and on the deployment side of this ledger, the early returns should concern American policymakers.
The Third Hand Theory is not a prediction that plumbers, electricians, mechanics, or construction workers will disappear. It is a framework for examining whether we have been measuring progress against the wrong benchmark. Rather than asking when robots will become fully equivalent to human workers, perhaps we should be asking how much value they need to add before businesses begin reorganizing work around their capabilities. Those are very different questions, and they may produce very different timelines.
Reading between the lines, the most significant story is not that robots are becoming more capable. It is that the machines already earning their keep in American factories and warehouses are doing exactly what the job site has always asked for—another set of hands—while public perception remains fixed on a science-fiction question about replacement that the market has already moved past. If the benchmark has quietly changed and the deployment gap with China is widening at the same time, then preparing workers, businesses, and educational institutions for this transition is less a matter of speculation and more a matter of prudent planning.
Disclaimer: This editorial reflects the analysis and opinions of the Craig Bushon Show Media Team based on publicly available reporting and industry developments. Deployment figures cited are drawn from company disclosures, securities filings, and industry research current as of July 2026. Forward-looking observations are necessarily uncertain and are offered to encourage thoughtful discussion rather than to predict specific outcomes. Readers are encouraged to review multiple sources and draw their own conclusions.
The Craig Bushon Show—we don’t just follow the headlines… we read between the lines to get to the bottom line of what’s really going on.
