What Happens When Chatbots Get a Body?

Over 3 million years, humans progressed from stone tools to the digital age. We have built, and destroyed, civilizations. Today, we can conjure up a video conversation with people from around the globe in the palm of our hand and chat with humans in outer space. Our mobile phones hold many times more information than what was contained in the Great Library of Alexandria. That same device is tens of millions of times faster than the computers that landed Apollo on the moon. As science fiction author Arthur C. Clarke predicted, our technology would appear like magic to generations past.

But what happens when those machines learn to beat us at our own games? When, for the first time in our history, humans have competitors that rival or surpass us in applied or, possibly even general, intelligence?

On February 10, 1996, thirty years ago, a computer took a step in that direction. That day marked the chess match between Russian grandmaster Garry Kasparov and IBM’s Deep Blue. Kasparov is among history’s greats, and he had beaten Deep Blue’s predecessor Deep Thought. In 1996, the grandmaster also won the match against Deep Blue, but it was close, and Deep Blue surprised many people by winning the first game. For the rematch a year later, Deep Blue was twice as powerful and able to calculate more than 200 million chess positions per second. In the rematch, Kasparov won only a single game out of six.

The last grandmaster to beat a computer was Ruslan Ponomariov in 2005. Today, humans can no longer compete with machines at chess, and computers play each other for championships. In 2016, Google’s AlphaZero taught itself to play chess in four hours. It then immediately dominated the day’s top-rated program (Stockfish 8) by making moves that, from a human perspective, were novel and often unorthodox—even sacrificing its queen. Grandmaster Peter Heine Nielsen remarked that he “always wondered how it would be if a superior species landed on Earth and showed us how they played chess, now I know.”

This world is now preparing for a future when a new class of machines dominates many of the tasks that have historically been the exclusive domain of humanity. We label this emerging tech Automata and define it as the integration of artificial intelligence with advanced robotics to create machines that can think, learn, and behave autonomously.

This leads us to ask important questions about how we respond to the rise of Automata: How do we adapt when the game we invented is being played by a mind that is alien and superior to our own? What does it mean for us to work with new and more advanced intelligent technology that can think and learn autonomously? How do we partner with machines that surpass our intelligence in many domains and may eventually possess superior general intelligence? What will happen to us?

In our view, as we move beyond chatbots, Automata will drive an unprecedented improvement in our standard of living. Economic history has been clear on this point: new technology tends to have a negative short-term effect on jobs, but in the longer run, new jobs are created, employment grows, and productivity increases to overall improve our lives.

We don’t, however, pretend that the transition to a better life will be easy. In fact, we expect it will be hard, and there is no guarantee that “better” will happen for everyone. Nevertheless, we are convinced that integrating AI with advanced robotics will revolutionize the way we live and work. The coming wave of technological change is a tsunami. Businesses and governments need to act quickly and with courage, because the transition is already underway. Traditional career paths are likely to disappear. We must learn to adapt, and to adapt again, as the pace of change continues to accelerate.

One of the most iconic scenes in cinema comes from the 1984 movie The Terminator, when Arnold Schwarzenegger walks into a police station, clad in leather and dark sunglasses, asking to see Sarah Connor. When he is told that he must wait, he says, in his thick Austrian accent: “I’ll be back.”

Forty years later, The Terminator continues to live large in public consciousness when we talk about the dangers of artificial general intelligence, or AGI. Schwarzenegger’s character was a terrifying, autonomous killing machine sent back in time to murder the mother of the man who will lead humans in the resistance (in the future) against intelligent machines. It wasn’t the plot that made the scene and the movie so iconic. What was particularly visceral about this Terminator was that it arrived in Schwarzenegger’s body, which made it easy to believe that he was an invincible assassin.

Without a physical form, even a super-intelligent machine will be limited in what it can accomplish. In fact, the growing fields of embodied cognition and embodied AI argue that to develop AGI, artificial brains need bodies. Without a physical form, AI will not be able to truly experience its environment, which will severely restrict its ability to learn and understand the world. If true, it is fortunate that progress in AI has been paralleled by advances in robotics, which will soon enable AI programs to operate in the physical world.

Automata—machines that can think, learn, and behave autonomously—is the difference between Deep Blue (a supercomputer that could beat humans at chess when specifically programmed to do so) and AlphaZero (a program that taught itself to play chess and other games at a level far beyond the capabilities of any human). This difference is moving the field toward machines like Smart Tissue Autonomous Robot (STAR) that plan and execute surgeries with only minimal supervision.

The development of general-purpose intelligent machines has sparked many start-ups (such as Figure and Prosper) to pursue innovations that allow Automata to autonomously carry out a variety of tasks. Agility, Boston Dynamics, Sanctuary AI, Apple, and Tesla are also active in building such machines. This approach moves away from a long history of highly specialized robots that are very capable because they are constrained to a specific job (such as the Canadarm or the Mars rovers) and toward Automata that are capable of a much larger and more diverse set of behaviours.

Nvidia has launched project GR00T, an AI platform designed to enhance the ability of robots to speak with people and emulate human movements in complex environments. GR00T can learn a task in a manner that is similar to the modelling process young humans and other animals use to acquire skills. Beyond Nvidia’s work, Tesla, Toyota, Google, Runway, and others are actively involved in developing general-purpose capabilities to power the next generation of Automata.

The rise of Automata will disrupt many different industries far more than the digital revolution and eliminate many types of work. At the same time, it will create demand for workers in roles that don’t yet exist. That transition will be very difficult for many people who grew up in a world where you could become an engineer, accountant, lawyer, dentist, doctor, professor, or any other number of jobs and expect a relatively stable career.

Moving forward, the human capital you accumulate through training and experience will become outdated at an ever-increasing rate. At the same time, new types of work will appear (and possibly disappear) faster than has been the case historically. Thriving in this new world of work will require people to be highly adaptable, creative, focused on accumulating transferable skills, comfortable with constant change, good at working with people, and willing to partner with smart machines.

If we extrapolate current trends ten or fifteen years into the future, it seems inevitable that Automata will begin to overtake humans in many domains. Intelligent machines are, or soon will be, cheaper, faster, and more precise. When these machines are available for service, they will be able to work nearly continuously and require only short breaks for repairs and upgrades—all without human needs, such as sleeping, eating, socializing, and leisure activities.

Will humans ever produce an AGI? Probably. Will it be in our lifetimes? Probably not. What we can confidently predict is that we have more pressing AI concerns to contend with—from privacy and human autonomy to misinformation and societal destabilization. Those problems are already here.

• Evan Solomon Wants Canada to Trust AI. Can We Trust Evan Solomon?
• Hello, I’m Phoenix. Would You Like Your House Cleaned?
• Your Job Is on Track to Becoming Economically Worthless

Although we do not expect the employment apocalypse that some have predicted, the reality is that there will be job losses. The impact will not be the same across all types of roles, and many future jobs will require partnerships between humans and Automata. Rather than trying to outperform AI on tasks that it is excellent or even superior at, humans should focus on higher-level strategic decision making, building relationships within and outside their organizations, and engaging in continual learning that facilitates rapid adaptation to the change that Automata will bring.

For that to be successful, organizations and governments will need to encourage lifelong learning and skill diversification. The focus should be on complementing rather than competing with the smart machines. We will need measures in place to mitigate prolonged unemployment, reduce the financial burden of mid-career retraining, and make shifting between types of work easier—that could, for example, include making it easier to transfer benefits and pensions between jobs. The transition will also likely require enhanced social protections for workers who are temporarily displaced as they move on to new types of work. Ideally, some of the financial benefits of this new technology will be committed to supporting the people made redundant by Automata.

As AI continues to advance, it can be our partner in the design of human work. Ultimately, we will want the agency to select our own path—we should not outsource who we are, or want to be, to smart machines. Our challenge will be to find the right balance between what we ask and allow machines to do and what we choose to protect and retain for humans.

Automata could affect the distribution of income and wealth. In the worst case, almost all the gains from Automata flow to a small percentage of the population who accumulate unprecedented wealth. That extreme inequality would undoubtedly result in severe social unrest. This risk is likely greatest in societies that allow the rise of Automata to be governed by market forces with minimal oversight and regulation. The responsibility for avoiding such a state will fall to national governments and economic unions that have the legislative power to set guardrails and use taxation to build infrastructure that facilitates broad participation in the new economy.

To intentionally design human work and avoid extreme inequality, governments will need to make some hard political and resource allocation decisions. Admittedly, we are concerned that governments focused on the immediate demands of voters will be unable to exercise the required delay of gratification. Yet, it has been done before: in times of war, America’s moonshot, national emergencies, nuclear weapons treaties, and in the creation of multinational trade agreements and economic unions. Avoiding extreme inequality will allow humanity to broadly benefit from the rise of intelligent machines in a more stable and secure world.

This means that planning for the future will require a constant cycle of setting strategy, measuring progress, and going back to the drawing board to rethink the strategy and how we will pursue it—keeping humans at the centre of important decisions.

Adapted and excerpted, with permission, from Automata: The Power of AI Integrated with Advanced Robotics by Ibrahim J. Gedeon and Kyle B. Murray, published by the University of Toronto Press, 2026.