Robotics, Machine Learning and the Emerging Skilled Workforce for Physical AI

The opening of Greenville Technical College’s new Center for Welding & Automation Excellence (CWAE) on August 20, 2026, attracted attention across the Upstate, including coverage by The Greenville News/Greenville Online and Greenville Journal, along with an official announcement from Greenville Technical College.

There is good reason for the attention. The approximately $32 million, 44,000-square-foot facility at Greenville Tech’s Brashier Campus near Simpsonville is described as the largest advanced welding facility in the Southeast. It expands the college from 90 to 122 welding stations and from four to 16 robotic welding stations, while adding specialized laboratories for robotic welding, pipe welding, laser welding, MIG welding, fabrication, nondestructive testing and what Greenville Tech describes as the first AI-powered welding laboratory in the United States. (Greenville Technical College⁠)

But that last description raises an important question for anyone working in robotics:

What does “AI welding” actually mean?

It does not primarily mean ChatGPT, large language models or generative AI. The artificial intelligence involved in advanced welding is much closer to the technologies at the heart of modern robotics: machine learning, computer vision, sensing, measurement, automated process control and intelligent robotic systems.

Understanding that distinction makes Greenville Tech’s investment considerably more interesting.

From Conventional Welding to Intelligent Welding Automation

Welding has always been a sophisticated feedback-control activity when performed by a skilled human.

An experienced welder observes the joint and weld pool, senses changes in the process, evaluates whether the weld is developing properly and continuously adjusts torch position, travel speed, heat input, electrode or wire feed and other parameters.

Automation attempts to transfer portions of that process to machines.

Conventional welding automation generally follows predetermined programs and control rules. A robotic system might move a torch along a programmed trajectory at a specified speed while maintaining predetermined electrical and mechanical parameters.

Intelligent welding adds another layer.

Cameras, position sensors, electrical measurements and other instrumentation can continuously observe what is happening. Software interprets those measurements. Machine-learning algorithms can recognize patterns in the resulting data. The system can then use that information to determine whether conditions are changing and, in advanced applications, adjust the welding process.

Conceptually, the architecture becomes:

Physical welding process → sensors and cameras → measurement data → machine vision/machine learning → process interpretation → control decision → robotic or welding-system adjustment → new measurements.
 That is a feedback loop remarkably similar to those increasingly appearing throughout advanced robotics.

This Is AI — But It Is Not an LLM

The broad term artificial intelligence now covers several substantially different technologies.

Generative AI systems and large language models such as ChatGPT are designed primarily to process and generate information represented as language, code, images and other digital forms. An LLM learns statistical relationships among tokens and uses those relationships to predict and generate sequences.

Industrial AI has a different job.

In welding, the important questions may be:

Where is the joint?

Where is the torch relative to the joint?

What is the geometry of the weld pool?

Is penetration or deposition developing as expected?

Has the workpiece shifted?

Are welding parameters drifting outside the desired operating envelope?

Should torch position, travel speed, wire feed or another parameter be changed?

Those are problems involving perception, measurement, estimation, prediction and control of a physical process.

The distinction can therefore be summarized simply:

Generative AI creates or interprets information. Industrial physical AI perceives, interprets and increasingly helps control the physical world.

There can eventually be considerable overlap between the two, but the welding technologies represented at Greenville Tech principally belong in the second category.

Machine Learning, Computer Vision and Measurement

It is also important not to describe every component of an automated welding system as artificial intelligence.

Measurement is not AI.

A voltage sensor measuring arc voltage is instrumentation. A current sensor measuring welding current is instrumentation. An encoder reporting robot position is instrumentation. A camera recording the weld pool is a sensing device.

The intelligence arises from what the system does with those measurements.

Machine learning allows software to learn relationships from examples and data rather than relying exclusively on manually programmed rules.

Computer vision applies computational methods, including machine-learning models, to images. In welding, that can allow a system to identify physical features such as joint geometry, torch position or characteristics of the weld pool.

Robotics supplies repeatable physical positioning and motion.

Process-control software determines how equipment should respond.

Nondestructive testing provides another source of information about whether the resulting weld meets requirements without destroying the component being inspected.

Put these capabilities together and the welding cell begins to acquire something analogous to a robotic perception-action loop:

sense → understand → decide → act → measure again.

For a robotics-oriented audience, that is probably a more useful way of thinking about AI welding than imagining an LLM controlling a welding torch.

A More Accurate Name for “AI Welding”
 For technical discussions, AI welding is probably too broad a term.

Depending upon the particular equipment and application, more informative terminology includes:

machine-learning-enabled welding automation, AI-enabled machine vision for welding, intelligent robotic welding, adaptive welding process control, or AI-enabled robotic welding automation.

These terms make clear that artificial intelligence is being incorporated into the perception and control architecture of an industrial process.

That distinction will become increasingly important as the term AI enters ordinary industrial vocabulary. A manufacturing plant may eventually contain dozens of forms of AI that have little resemblance to a conversational chatbot.

Building the Workforce Around the Technology

The technology matters because Greenville Tech is responding to a workforce problem, not conducting an academic AI experiment.

At the center’s opening, Greenville Tech President Larry Miller said there are approximately 400,000 welding-technician job openings nationally and that upcoming South Carolina energy projects alone are expected to require approximately 5,400 welders. He also described a serious replacement imbalance: for approximately every four welders retiring or otherwise leaving the occupation, only one new welder is prepared to enter. (Greenville Journal⁠)

At the same time, the definition of a welder is expanding.

Greenville Tech’s own Welding and Robotic Technology Certificate combines conventional welding instruction with courses including Robotic Welding I and Robotic Welding and Manufacturing. The college describes the program as preparing graduates to provide skilled support in specialized and robotic welding for manufacturing and construction. (Greenville Tech Student Handbook⁠)

Tomorrow’s advanced welder may therefore need to understand not only how to produce a high-quality weld manually but also how to work with robotic welding cells, computerized equipment, machine-vision systems, process data and automated quality systems.

That does not necessarily eliminate the skilled tradesperson. It can make the skilled tradesperson more technologically capable and potentially more valuable.

Where These Skills Will Be Needed

The potential employment base is broad.

Advanced manufacturing requires welding for machinery, production equipment and fabricated assemblies. Automotive and transportation manufacturing rely heavily on robotic joining and increasingly sophisticated automation. Aerospace and defense applications demand high precision and extensive quality control. Construction and structural fabrication require enormous quantities of welding expertise.

Energy infrastructure represents another major opportunity. Power-generation projects, pipelines, pressure systems and industrial plants can require highly specialized welding and inspection. The growth of data centers and the electrical infrastructure needed to support them creates additional demand throughout the industrial construction supply chain.

Shipbuilding, heavy equipment, metal fabrication, plant maintenance and repair operations create still more applications.

The result is an unusual labor-market combination: industry needs more welders at the same time that it needs increasingly sophisticated welders.

Greenville Tech is attempting to address both requirements.

How the Center Is Organized

The physical facility reflects that objective.

The 44,000-square-foot center contains 122 welding stations overall and 16 robotic welding stations, along with instructional laboratories covering robotic welding, pipe welding, laser welding, MIG welding, fabrication, nondestructive testing and AI-enabled welding. Classrooms and the primary welding facilities occupy the ground floor. (Greenville Journal⁠)

Earlier Greenville Tech procurement documents described the project as a two-story advanced welding facility incorporating classrooms, offices, general welding booths, robotic welding booths, simulation welding booths and an exterior cutting and grinding area. (Greenville Technical College⁠)

The result is closer to an advanced manufacturing learning environment than a traditional welding shop.

Students can encounter the progression from conventional manual skills through automation, robotics, sensing and advanced process technology within the same educational environment.

Extending the Pipeline Into High School

The project also has implications beyond Greenville Tech itself.

Because Greenville Tech is moving its welding operations into the new center, space previously used by the college can support an expanded Greenville County Schools welding program. Greenville County Schools and Greenville Tech formalized that partnership earlier this year. (Greenville Journal⁠)

That creates the beginnings of a regional talent pipeline:

high-school technical education → Greenville Technical College → advanced welding and robotics education → Upstate manufacturing, construction and energy employers.

That pipeline may prove every bit as important as the equipment inside the building.

A Public-Private Investment in Industrial Skills

The approximately $32 million facility was assembled through a combination of public investment and private-sector participation.

Greenville Tech reports that the South Carolina General Assembly provided $15 million and $3 million in federal funding was secured for the project. Industry, foundations and other partners supplied additional financial support, equipment and technical expertise. (Greenville Technical College⁠)

That collaboration is important because industrial technology evolves much faster than a technical curriculum can if education and industry operate independently.

Employers know which equipment, processes and skills are appearing in factories. Technical colleges know how to turn those requirements into education and credentials. Government can help provide the capital infrastructure that would be difficult to duplicate independently.

CWAE brings those three pieces together.

Physical AI Comes to the Skilled Trades

The broader significance of the Greenville Tech center extends well beyond welding.

For the last several years, public understanding of artificial intelligence has been dominated by generative AI. That is understandable because ChatGPT and similar systems provide millions of people with their first direct interaction with a powerful AI system.

But another AI transformation is occurring simultaneously.

AI is moving into machines.

Machine vision allows machines to perceive. Machine learning allows systems to recognize patterns and make predictions. Sensors connect computation to physical conditions. Robotics allows software to produce physical action. Feedback control closes the loop.

This convergence is increasingly being described as physical AI: intelligent computational systems interacting directly with the physical world.

A sophisticated robotic welding cell is an excellent example.

The AI is not sitting behind a screen writing an essay. It is helping a machine understand what is physically happening and potentially determine what should happen next.

That is an important concept for manufacturing, robotics and workforce education—and Greenville Tech is putting it directly into the classroom.

Automation Can Increase the Need for Skilled People

There is an understandable tendency to frame robotics as a simple substitution problem: install a robot and eliminate a worker.

Welding demonstrates why reality can be more complicated.

If industry already faces a shortage of qualified welders, robotic welding can allow skilled people to supervise, program and support automated systems while concentrating human expertise where it produces the greatest value.

But those systems themselves require people.

Someone must understand welding metallurgy and technique. Someone must configure the robotic cell. Someone must troubleshoot the process. Someone must interpret quality information. Someone must understand what the sensors are reporting. Someone must determine whether automation is producing an acceptable weld.

Consequently, robotics can change the skill composition of welding employment rather than simply eliminating welding employment.

That makes education essential.

A Significant Opportunity for Upstate South Carolina

For Upstate South Carolina, CWAE represents a particularly valuable capability because it sits in one of the country’s important advanced-manufacturing regions.

It provides employers with a pipeline of workers trained not only in established welding processes but increasingly in the intersection of welding, robotics, automation, sensing and intelligent manufacturing.

It provides Greenville Tech with an environment where curricula can evolve alongside industrial technology.

And it provides young people throughout the Upstate with another pathway into technically sophisticated careers that do not necessarily require beginning with a traditional four-year university degree.

That last point deserves attention.

Modern manufacturing careers increasingly blur the old distinction between a “trade” and a “technology career.” A young man or woman learning welding today can potentially progress from manual welding to robotic welding, automation, inspection, programming, supervision, manufacturing engineering support or specialized applications in energy, aerospace and advanced manufacturing.

Experienced welders in specialized processes can earn substantial incomes, although actual compensation varies considerably by certification, specialty, industry, experience, overtime and travel requirements. The important point is not that every welding graduate is guaranteed a particular salary. It is that advanced welding provides a credible pathway into a skilled occupation from which increasingly sophisticated technical careers can develop.

For a young person beginning a career and eventually supporting a family, that can be a meaningful economic opportunity.

The Larger Lesson

The Center for Welding & Automation Excellence illustrates something that is likely to become increasingly visible throughout American industry.

Artificial intelligence is leaving the computer screen and entering the physical workplace.

In Greenville, that transition can be seen in a welding booth.

Sensors measure the physical process. Cameras observe it. Machine-learning systems can interpret what they see. Robotic equipment executes precise movements. Adaptive control can modify the process. Skilled humans understand, supervise, troubleshoot and improve the overall system.

That is not the AI of a chatbot.

It is physical AI applied to manufacturing.

And perhaps the most significant aspect of Greenville Tech’s new center is that it does not treat the skilled worker and artificial intelligence as opposing forces. It is preparing people to work with increasingly intelligent machines.

As American manufacturers, construction companies, energy projects and other industries search for thousands of additional welders, Greenville Technical College is making a substantial investment in the proposition that the future welder will remain a skilled tradesperson—but increasingly a skilled tradesperson who understands robotics, automation, measurement and artificial intelligence.

That combination could make the Center for Welding & Automation Excellence important not only to Greenville Technical College, but to the future industrial workforce of Upstate South Carolina. AI welding.