The Amazing Advancements of Robotics: A Long Road Ahead for Society

October 2026 Update: The Robots Are Learning, but Is Society Keeping Up?

Editor's note — October 2026: This article was originally written in November 2024. The original observations have been preserved to illustrate how rapidly robotics has developed. A new update explores the significant technological advances and ongoing societal questions.

When I originally wrote this article in November 2024, I was fascinated by a Boston Dynamics robot moving components around a factory. It was an extraordinary demonstration of engineering, but what struck me most was not what the robot could do. It was what it couldn't do.

A human watching the same task would probably have found a more efficient way to complete it within minutes. The robot could perform the movements, but it lacked the broader understanding and initiative that humans take for granted.

Almost two years later, robotics has advanced considerably, and some of the limitations I discussed are beginning to disappear.

Unfortunately, I am less convinced that society has made the same progress in preparing for the consequences.

From Impressive Demonstrations to Actual Employment

One of the biggest developments since 2024 is that humanoid robots are beginning to move beyond carefully controlled demonstrations and into real industrial environments.

In 2025, BMW worked with robotics company Figure AI to introduce its Figure 02 humanoid robot into production at its Spartanburg factory in the United States. They have now announced the use of Figure 03.

During the deployment, the robots handled more than 90,000 components, accumulated approximately 1,250 operating hours and contributed to the production of over 30,000 BMW vehicles.

This wasn't simply a robot performing a clever trick for a promotional video. It was a robot undertaking repetitive physical work in an actual manufacturing environment.

By 2026, BMW had progressed to testing Figure 03 on more complicated logistics tasks and had introduced another humanoid robotics pilot at its Leipzig factory in Germany.

Meanwhile, Boston Dynamics unveiled a production version of its electric Atlas humanoid robot in January 2026, with initial deployments planned for Hyundai and Google DeepMind.

These developments do not mean that factories are about to become completely automated or that humanoid robots can replace every worker. They are still early deployments, often involving tightly defined tasks and carefully managed environments.

Nevertheless, the transition from laboratory demonstrations to productive employment is significant.

The question is no longer simply whether a robot can perform human work.

It is increasingly becoming a question of how much work it can perform, how reliably it can perform it and whether it can do so economically.

The Robot Is Beginning to Understand the Task

Perhaps the most important advancement is not the robot's ability to walk, lift objects or maintain its balance.

It is what is happening inside its artificial intelligence.

In 2024, I suggested that a robot moving components between two racks would be unlikely to recognise that moving the racks closer together might improve efficiency.

Traditional industrial robots are exceptionally good at repeating precisely defined operations, but they generally have little understanding of the wider purpose of their work.

That is beginning to change.

Google DeepMind's Gemini Robotics research has demonstrated AI systems that combine visual perception, language understanding, spatial reasoning and physical actions.

Rather than relying entirely on a fixed sequence of instructions, these systems can interpret their surroundings, respond to natural-language instructions and adapt their behaviour when circumstances change.

Research has also demonstrated robots performing unfamiliar variations of tasks and manipulating objects in ways that were not individually programmed in advance.

This is a substantial step forward.

However, we should be careful not to confuse adapting to a situation with possessing a human understanding of it.

A robot might recognise an object, work out how to grasp it and place it in the correct position. That doesn't necessarily mean it understands why the work is being undertaken, whether the task is worthwhile or whether reorganising the entire workspace would be a better solution.

And there is another distinction worth making: a robot using previously trained knowledge to handle a new situation is not necessarily learning permanently from every experience.

Some systems are becoming increasingly adaptable, but genuine long-term autonomous learning in unpredictable environments remains an enormous challenge.

Returning to my original factory example, I suspect that it is becoming less difficult to build a robot that can move parts between racks.

The more interesting challenge is developing a robot that independently decides the racks are in the wrong place.

Five Million Robots Already at Work

While humanoid robots attract most of the headlines, the much larger transformation is already taking place with conventional industrial robotics.

According to the International Federation of Robotics, approximately 5.08 million industrial robots were operating in factories around the world in 2025, with more than 600,000 new installations during that year alone.

China accounted for approximately 59% of worldwide new industrial robot installations in 2025.

These figures are remarkable, but they also remind us that the robotics revolution is not entirely dependent on humanoid machines.

A robot does not need a face, arms and two legs to replace work previously carried out by a person.

In many circumstances, a specialised robotic arm, automated production line or autonomous warehouse vehicle will be cheaper, faster and more reliable than a machine designed to resemble a human.

Humanoid robots may eventually become particularly useful in environments originally designed for people. But the economic transformation caused by robotics has already been underway for decades.

And now we are adding increasingly capable AI to those machines.

The Missing Ingredient: Common Sense

Despite the progress, I believe robotics still faces an extraordinary challenge.

The real world is messy.

Objects are dropped. Furniture gets moved. People walk into unexpected places. Tools break. Instructions are incomplete. Something that worked perfectly yesterday suddenly stops working today.

Humans deal with these problems using a lifetime of accumulated experience, often without consciously thinking about what they are doing.

A robot might successfully fold a shirt in a laboratory, but how will it cope when the shirt is tangled around another garment, the laundry basket has been moved and the family dog decides to investigate?

This is why impressive videos should not be confused with demonstrations of unlimited general-purpose intelligence.

A machine successfully completing a task once is very different from a machine completing thousands of different tasks reliably, safely and without human assistance.

Nevertheless, improvements in visual reasoning, physical manipulation and AI models capable of controlling robots are gradually reducing this gap.

We may still be a considerable distance from a genuinely general-purpose household robot, but the direction of development is becoming clearer.

The Economic Question We Still Haven't Answered

This brings me back to the concern that motivated the original article.

What happens to a society when an increasing proportion of its work can be performed by machines?

Historically, technological development has destroyed certain occupations while creating entirely new ones.

The industrial revolution transformed agriculture and manufacturing. Computers eliminated many clerical tasks while creating enormous new industries.

It is possible that AI and robotics will follow a similar pattern.

But I am not convinced we should simply assume that they will.

For the first time, we are developing technologies capable of automating both physical work and significant elements of intellectual work at the same time.

A factory robot can handle components. AI software can manage paperwork, process orders, answer customers and assist with complex decisions.

Combined, they could fundamentally change how businesses operate.

Consider a manufacturing company that currently employs 100 people.

If robotics and AI eventually allow it to produce the same amount with 30 employees, the business could become substantially more competitive.

But what happens to the other 70 people?

Some might find new employment. Others might retrain. New industries might emerge.

However, there is no guarantee that suitable new jobs will appear in the same locations, at the same salaries or quickly enough to prevent considerable hardship.

And this raises a broader economic contradiction.

Businesses benefit from reducing employment costs, but businesses also depend upon consumers having money to purchase their products.

If automation increases production while reducing the purchasing power of a significant proportion of society, who is going to buy everything we produce?

This is not an argument against robotics. It is an argument for recognising that technological progress and economic prosperity are not automatically the same thing for everyone.

Who Benefits From the Productivity?

We often discuss robotics in terms of jobs lost or jobs created.

Perhaps we should be asking a different question.

If machines allow society to produce more goods and services with less human effort, why shouldn't everyone benefit from that achievement?

In theory, greater productivity could mean shorter working weeks, improved public services, lower costs and more time for people to enjoy their lives.

That sounds like progress.

But there is another possible outcome: enormous productivity gains concentrated among a relatively small number of businesses and investors, while increasing numbers of people struggle to find secure employment.

Neither outcome is inevitable.

The difference will depend upon the economic and political decisions we make as these technologies develop.

Universal Basic Income remains one possible response, although it raises difficult questions about affordability, incentives and the purpose that employment provides for many people.

Other approaches might include sharing productivity gains through shorter working hours, better retraining, wider employee ownership and changes to taxation that reflect an increasingly automated economy.

There is unlikely to be one simple answer.

What worries me is that these discussions are still far less prominent than the extraordinary technological developments that make them necessary.

The Next Two Years Could Be Even More Interesting

Looking back at what I wrote in 2024, I would change one important observation.

I originally suggested that robots could not learn or adapt without explicit programming or human intervention.

That is no longer a fair description of the most advanced experimental systems. AI-powered robots are demonstrating increasing abilities to generalise from training, interpret unfamiliar situations and adjust their actions.

What they have not yet demonstrated is the full range of flexibility, judgement, independence and reliability that we expect from human workers.

That distinction matters.

But I suspect that the gap will continue to narrow, and possibly more quickly than many people expect.

The next major breakthrough may not be a robot performing an extraordinary backflip or dancing at a technology exhibition.

It may be something far less spectacular.

A robot that looks at a problem, recognises that the existing method is inefficient and independently discovers a better way of doing the job.

And when that happens consistently, across many different environments, we will have crossed another important threshold.

Final Thoughts: Technology Is Advancing. Are We?

I remain enormously enthusiastic about robotics and artificial intelligence.

These technologies could make dangerous occupations safer, reduce exhausting physical work, support elderly and disabled people and improve productivity in ways previous generations could barely imagine.

But there is a distinction between creating technology that can do remarkable things and creating a society that benefits from those remarkable things.

We are investing billions in developing increasingly capable machines.

Perhaps it is time we invested a little more thought into what happens to the people those machines are designed to help — or potentially replace.

In 2024, I concluded that we had a long road ahead.

In October 2026, I would argue that the robots have travelled a remarkable distance along that road.

I'm just not sure the rest of us have started walking yet.

---

AI Robotics on YouTube

---

Original Article published in November 2024

Robotics and AI are advancing at an incredible pace, transforming industries and reshaping the workforce. The latest demonstration by Boston Dynamics showcases a humanoid robot capable of independently performing tasks like moving car parts in a factory setting. The robot selects a part from one rack and moves it to another while also dealing with errors or disruptions along the way—truly impressive feats that highlight the amazing progress in robotics and AI.

Yet, while this demonstration is an incredible step forward, it also reveals just how far we still have to go to match the ingenuity and adaptability of human workers.

The Strengths and Limitations of Current Robotics

The Boston Dynamics video is nothing short of a marvel. The robot efficiently moves between racks, carefully selects a part, and transfers it to the target location. The robot can deal with minor issues, such as tripping, and continues its task seamlessly. In a factory setting, these actions showcase how robotics can replace repetitive, mundane manual tasks.

However, if we look closer at what makes humans so exceptional, we can see the limitations of current robotic technology. A human worker would quickly adapt by moving the second rack closer, minimizing movement and saving time. A person would also likely figure out that they could carry more than one part at a time, further doubling their efficiency. These kinds of adaptations are second nature to humans but are not yet part of robotic capabilities without explicit programming or human intervention.

For now, these robots cannot learn on the fly—they do what they are programmed to do. The entire process relies on a supervisor to notice inefficiencies, which highlights that we are far from the dream of robots matching the full cognitive ability of a human worker.

Human Workers vs. Automation

Of course, it's easy to get caught up in the amazing advancements in robotics. After all, the capabilities demonstrated by Boston Dynamics are light-years beyond what was possible just a decade ago. Robots don’t get tired, they can work 24 hours a day, and companies are eager to avoid the costs associated with labor laws, breaks, and salaries.

However, these advancements also raise fundamental questions about the very nature of employment and society. A considerable portion of jobs—whether in factories or offices—is made up of repetitive tasks. For instance, my experience working in a factory involved stamping small plastic disks with unique serial numbers. While the task was simple, boredom would inevitably set in after hours of repetitive actions, and mistakes were unavoidable. It’s precisely these kinds of tasks that automation can do far more accurately, without getting tired or losing focus.

But therein lies the dilemma: if such jobs didn’t exist, millions of people wouldn’t have a way to earn a living. Automation offers tremendous benefits, but it also presents societal challenges that we must address if we want to avoid severe repercussions.

The Societal Dilemma of Automation

The societal implications of widespread automation are profound. Automation and AI are reducing the need for human labor, and governments are already feeling the effects of a shifting employment landscape. Lower-income tax revenues and increasing demands for social support programs are becoming pressing issues. The problem cannot be solved by simply increasing taxes on those still employed—doing so will likely only hasten the shift toward more automation, leading to fewer jobs and less revenue in an accelerating downward spiral.

The harsh reality is that governments can’t fund both an aging population and mass unemployment. The more people lose their jobs to machines, the more they lose their ability to spend money in the economy. And as history has shown us, when people are hungry and desperate, civil unrest follows. The true threat to societal stability may not come from international conflicts but from within, as the gap between those with access to opportunities and those without continues to grow.

The Need for a New Approach

For decades, we've heard about the promises of a four-day workweek, automation, and increased leisure time. Yet, despite these promises, the majority of workers are facing longer hours and more insecurity. Part of this is due to the fact that past technological advancements created jobs—factories needed workers, logistics required staff, and new gadgets needed people to sell them.

But now we are witnessing a technological revolution that, for the first time, seems to be reducing the need for human labor on a broad scale. Factories today require fewer and fewer workers. Call centers are being replaced by AI chatbots, and more people are making purchases online, eliminating retail positions. These rapid changes will soon bring about massive societal shifts that we’re not yet ready to handle.

The Case for Universal Basic Income (UBI)

One proposed solution is a Universal Basic Income (UBI), guaranteeing citizens a minimum income regardless of employment status. In theory, this could prevent people from falling into poverty as automation takes over jobs. However, this too has its limitations. A society that relies entirely on UBI risks becoming stagnant—where the majority are dependent on state support, while only a select few thrive by setting up businesses or leveraging their unique skills. UBI might ensure that people can survive, but it does little to ensure people thrive, find purpose, or feel fulfilled.

The Importance of Proactive Policy and Education

Another critical component of dealing with the rise of automation is education. Not everyone has the inclination or skills to become an entrepreneur. Not everyone wants to learn coding or AI—nor should they have to. Governments need to implement policies that prepare the workforce for this new reality. This includes upskilling and reskilling workers to take on roles that cannot be easily automated, such as those requiring creative problem-solving, empathy, and complex decision-making.

Facing the Future

We’re at a crossroads. While the world focuses on conflicts in Ukraine, North Korea, or the Middle East, a quieter yet more dangerous threat is brewing—a systemic issue that could lead to mass unemployment, social instability, and the collapse of economic systems if not handled correctly. Automation is here, and it’s moving faster than we expected. We can no longer afford to kick the problem down the road.

The incredible advancements in robotics and AI are certainly paving the way for a future of unprecedented efficiency, but they are also paving the way for a societal revolution. Whether this leads to greater prosperity or greater unrest will depend on how we, as a society, choose to address these changes.

Our modern world is fragile. It’s time we start talking about these issues openly and bring them to the forefront of politics and public debate. Robotics and automation have immense potential to improve our quality of life, but only if we’re prepared to adapt our systems and our mindsets to accommodate them. Otherwise, the road to progress might lead us down a dangerous path, paved, as always, with good intentions.

---

Latest Future AI Articles