AI Is the Tool. We Are the Toolsmen.

Human adaptation, cognitive redundancy, and the expanding frontier of human ambition.

1. We Have Never Needed to Be the Machine

A crane can lift loads no human can lift. An aircraft can travel faster than a human body ever could. A calculator can perform arithmetic faster and more accurately than most of us. Computers can store and process information on a scale that exceeds our unaided abilities.

Yet we continue to use these machines. Their superior capabilities have not made them useless to us. In many cases, those capabilities are precisely what make them valuable.

This seems obvious when we consider physical machines. Nobody expects a construction worker to lift more than a crane or a pilot to fly faster than an aircraft. We build machines to overcome our limitations, then learn to operate them.

Why should the principle suddenly become invalid when the machine begins to perform intellectual work?

Artificial intelligence challenges our traditional understanding of human capability because it can perform tasks that we have long regarded as distinctively intellectual. It can calculate, write, analyse, design, and solve problems. Some of these capabilities already exceed those of many individual humans in particular contexts, and the range continues to expand.

But perhaps we are asking the wrong question when we ask whether AI will become more intelligent than us.

A more useful question is this: What can human beings accomplish when they learn to use intelligence that they could never produce independently?

2. The Calculator Did Not End Mathematics

Before electronic calculators became commonplace, routine arithmetic demanded considerable human effort. People performed calculations by hand, memorised procedures, and developed numerical fluency through repeated practice.

Calculators changed the economics of that effort. Many calculations that once required considerable time could be completed almost instantly.

We did not need to become better at mental arithmetic to benefit from this change. We needed to learn how to use the calculator, understand which operations to perform, and interpret the results.

Something else happened, too. For many people, routine arithmetic became less necessary in everyday work. Some of the skills that earlier generations practised extensively were no longer exercised as frequently. This was a real trade-off, not an illusion.

Yet the availability of calculators also made more elaborate calculations accessible. People could spend less time executing routine operations and more time exploring problems in which those operations were only one component.

The calculator did not automatically make every person a better mathematician. Nor did it make mathematical understanding unnecessary. But it changed the relationship between human effort and mathematical capability.

The important distinction is between the skill we no longer need to perform ourselves and the larger activity that the skill once supported.

A task can become redundant without the broader human pursuit becoming redundant.

This distinction matters enormously as AI begins to perform cognitive tasks that previously demanded years of training.

3. The Inventor and the User Live in Different Worlds

Consider a car. A ten-year-old, given suitable instruction and assuming the appropriate legal and safety conditions, can learn the basic principles of operating one. The child does not need to understand ignition systems, combustion, fuel injection, transmission engineering, or electronic engine management.

The engineers who developed the car had to solve problems that the driver never encounters directly. Their knowledge is embedded in the machine, and the driver inherits the benefits of that work.

The same principle applies to a pilot.

A pilot must understand aircraft performance, navigation, flight controls, operating procedures, and emergency response. But the pilot does not need to have designed the engines, manufactured the turbine blades, developed the avionics, or written the flight-control software.

The pilot’s expertise is real, but it is different from the expertise required to build the aircraft.

We can distinguish three activities: inventing a capability, understanding how it works, and applying it to achieve a result. These activities overlap, but they are not interchangeable. Application can require substantially less knowledge than invention, although effective and safe application still requires appropriate training and judgment.

Modern technology is full of such asymmetries. No individual understands every component of the systems on which modern civilisation depends. We use electricity, telecommunications, navigation systems, medical equipment, and computer networks without mastering all the science and engineering behind them.

This is not an accidental weakness of modern society. It is one of the ways complex civilisation becomes possible.

The knowledge required to create a capability does not have to be reproduced in every person who uses it.

4. Cognitive Redundancy Is Not Human Redundancy

Now imagine that AI can perform a growing range of intellectual tasks more efficiently than people can perform them unaided.

A student might use AI to explore a mathematical problem that would otherwise be beyond their present abilities. A researcher might investigate possibilities that would take months to examine manually. A programmer might describe a desired application and use AI to generate and test much of its implementation.

In each case, the person may accomplish something without independently possessing every skill that the task would previously have required.

Some traditional cognitive functions may consequently become less necessary for particular purposes. Routine calculations, memorised procedures, standard code production, and certain forms of analysis may increasingly be delegated to machines.

We should not pretend that nothing will be lost. When people stop practising a skill, their proficiency can decline. When a machine performs a task, dependence on that machine can grow. And when people cannot evaluate an answer, they may struggle to recognise when the system is wrong.

Nevertheless, losing the need to perform a particular mental operation is not equivalent to losing the ability to pursue the larger objective.

We do not require every pilot to become an aircraft engineer. We do not require every calculator user to become an expert in electronic circuits. We do not require every person using a computer to understand its operating system at the level of its designers.

Similarly, we may not need every person working with AI to reproduce all the reasoning, coding, analysis, or knowledge-generation capabilities that AI makes available.

The important question becomes which capabilities we must retain ourselves, which we can safely delegate, and which new abilities we must develop to operate the resulting systems effectively.

Technology can make some human skills less necessary while making human achievement more powerful.

5. The Ferrari Principle

Imagine giving someone access to a Ferrari.

The driver does not become capable of running at the car’s maximum speed. The driver does not acquire the engineering knowledge required to build its engine. The driver’s muscles do not become stronger, and their biological limitations remain unchanged.

Yet the person can travel at speeds and over distances that would otherwise be impossible for them.

The machine changes the person’s effective capabilities without fundamentally changing the person’s biology.

AI may produce a comparable transformation in intellectual work. An individual who cannot independently write a sophisticated software system might use AI to build one. Someone without specialist training might investigate a scientific question with the assistance of AI tools. A small team might accomplish work that previously required a much larger organisation.

The person has not necessarily acquired all the underlying expertise. Instead, the person’s effective capabilities have expanded through access to a more powerful instrument.

This is why the distinction between individual intelligence and effective capability matters.

We often evaluate people by asking what they can do alone. But human civilisation has always depended on what people can accomplish collectively through tools, institutions, accumulated knowledge, and cooperation.

AI may push this principle into a new domain. The tools we use will not merely extend our physical strength or computational capacity. They will increasingly participate in the intellectual processes through which we create, reason, and solve problems.

The Ferrari is not the destination. It is the means by which we reach destinations that were previously inaccessible.

6. Every New Plateau Opens Another Horizon

Human beings adapt, but adaptation does not mean that every old skill survives or that every individual benefits equally. It means that people can develop new ways of living and working when their circumstances change.

Cars reduced the need to walk long distances. Calculators reduced the need to perform many routine calculations manually. Computers changed how we store information, communicate, and carry out complex work.

Each technology changed the relationship between human effort and what could be achieved.

As one limitation weakened, other possibilities became practical. Faster transportation made new forms of trade and settlement possible. Computing enabled scientific investigations that would have been prohibitively expensive or time-consuming. Digital communication created opportunities that did not exist in the same form before.

This suggests a recurring pattern: when a constraint is removed, the frontier of possible achievement can move outward.

We can call this the New Plateau Principle.

A plateau represents a level of capability that becomes achievable after a previous limitation is overcome. Once that level becomes ordinary, it no longer defines the ultimate boundary of ambition. People begin pursuing what lies beyond it.

AI could create a similar transition. If machines take over more routine intellectual work, the next plateau might involve coordinating complex systems, investigating more ambitious scientific questions, creating new technologies, or exploring problems that currently exceed our practical capacity.

And our goals need not remain limited to immediate improvements in convenience or productivity.

Human ambition already extends toward space exploration, understanding the universe, radical extensions of healthy human lifespan, and possibilities we cannot yet describe clearly. These ambitions may prove difficult or impossible in some forms, but they illustrate how the horizon of human goals can extend beyond the constraints of the present.

The removal of a limitation does not dictate what people will want next. It creates possibilities among which they can choose.

That distinction is important. A new plateau is an opportunity, not a guaranteed destination.

7. AI May Help Us Adapt to AI

Every powerful technology introduces new problems alongside its benefits. Industrialisation transformed production but also created serious social and environmental challenges. Motor vehicles expanded mobility while creating traffic risks, pollution, and new demands on infrastructure.

Human societies developed ways to manage some of these problems, although many required decades of effort and some remain unresolved.

AI will also create challenges. These may include changes in employment, misinformation, security risks, concentration of power, overdependence on automated systems, and failures that are difficult to detect.

But AI may also help us understand and address the challenges it creates.

We could use AI to detect fraudulent AI-generated content, test other AI systems, identify vulnerabilities, investigate the consequences of automation, and develop methods for evaluating model behaviour. We could use AI-assisted research to accelerate the search for solutions to technical and social problems.

This creates the possibility of recursive adaptation: using a technology’s capabilities to respond to the difficulties introduced by that same technology.

The process is not guaranteed to succeed. AI systems can make mistakes, amplify existing problems, and produce solutions that create further complications. Some challenges require political decisions, institutional reform, human accountability, and values that cannot be settled by technical optimisation alone.

Nevertheless, we should not assume that humanity must confront every AI-related problem using unaided human intelligence. AI itself may become part of our adaptive capacity.

The tool may help us navigate the world that emerges after the tool changes it.

8. AI Is the Tool. We Are the Toolsmen.

The history of technology suggests that human progress does not require us to remain superior at every task we delegate to machines.

We do not need to be stronger than a crane, faster than an aircraft, or better at routine arithmetic than a calculator. We need to understand what these tools can accomplish, how to use them effectively, and where their limitations matter.

The same principle may increasingly apply to intelligence.

The human role will not necessarily remain fixed. AI may assist with planning, evaluation, and even the formulation of goals. Some occupations may disappear, some will change, and new opportunities may emerge. The transition will require people and institutions to adapt, and there is no guarantee that the benefits will be distributed fairly.

But the possibility of machines outperforming humans in particular intellectual tasks does not, by itself, establish that humans will have no useful role. It changes the question of where that role lies.

Perhaps expertise will increasingly involve specifying objectives, choosing appropriate tools, coordinating their work, evaluating their results, and deciding which problems deserve attention. Perhaps we will place greater value on judgment, responsibility, direction, and the ability to identify worthwhile goals.

These functions may also be assisted by AI, so they should not be treated as permanently exclusive human privileges. Our task will be to determine how to use increasingly capable systems while retaining meaningful control over the outcomes we care about.

The central idea is simple: we have repeatedly expanded our capabilities by building tools that exceed our natural abilities. We have then learned to operate those tools without reproducing all the expertise embedded in them.

AI may extend that pattern from the physical world into the world of thought.

The carpenter is not defined by the strength of the hammer. The pilot is not defined by the power of the engine. And perhaps the human mind will not be defined solely by what it can accomplish without assistance, but also by the worlds it can reach through the tools it learns to use.

The question is no longer simply how intelligent our tools will become. It is what humanity will choose to do with the capabilities those tools make possible—and whether our capacity to adapt will keep pace with the power we create.

Continue reading

Explore the related article The Last Thing Humans Should Outsource to AI for a deeper look at the cognitive abilities we may need to preserve as AI becomes more capable.