The question is no longer just what technology can do for us. It is what we want technology to make possible.
We have become remarkably good at asking what technology can do. Artificial intelligence can write reports, analyse information, answer questions and support decisions that previously required considerable human effort. Algorithms can process enormous amounts of data, identify patterns and perform tasks with a speed and consistency that people cannot match. As these capabilities improve, activities that once seemed impossible to automate are increasingly becoming candidates for technological replacement.
For businesses and public organisations, the opportunities are substantial. Automation can reduce costs, increase productivity and make services available to more people. It can eliminate repetitive work, improve access to information and free employees to concentrate on activities that require human expertise. For individuals, technological assistance can create greater independence and make tasks possible that would otherwise be difficult or inaccessible.
Yet somewhere between discovering that an activity can be automated and deciding that it should be automated, an important question is often overlooked: What is the activity actually for?
We tend to think of work and everyday activities in terms of the outcomes they produce. A report communicates information, a conversation resolves a question, and an educational assignment demonstrates knowledge. If technology can produce the same outcome more efficiently, replacing the human activity may appear to be an obvious improvement.
But activities can create value beyond their immediate results. The process of preparing a report may develop professional judgement. A conversation may build trust and understanding. An assignment may help a student learn to reason, evaluate evidence and communicate ideas. When we automate these activities, we may remove more than the effort required to complete them.
This does not mean that automation is inherently undesirable or that human involvement should always be preserved. It means that decisions about automation require a broader understanding of what technology changes, including the capabilities, relationships and opportunities that may be strengthened or weakened in the process.
When efficiency becomes the objective
Efficiency has become one of the most influential measures of technological progress. If a machine can perform an activity faster, more accurately or at a lower cost, automation offers an apparently straightforward benefit. In organisations facing financial constraints, labour shortages or growing demands, such improvements can be essential.
There are good reasons to welcome many forms of automation. Repetitive administrative tasks consume time that could be spent on more meaningful activities. Automated systems can make services available outside conventional opening hours and help people overcome barriers that previously limited their independence. For professionals working under considerable pressure, reducing routine tasks may create opportunities to concentrate on complex decisions, personal interactions and specialist work.
The difficulty arises when efficiency becomes the primary measure of success rather than one consideration among several. A process may become faster without becoming better. An organisation may reduce the resources required to deliver a service while transferring additional work to the people using it. A professional may complete more tasks while having fewer opportunities to develop the expertise needed to perform them independently.
Efficiency is not a neutral objective. It reflects decisions about which outcomes matter and which forms of value deserve attention. When a process is evaluated primarily through its cost, speed and measurable output, other qualities can become difficult to recognise. Trust, learning, judgement, participation and confidence may be essential to the value of an activity even when they cannot easily be expressed in financial terms.
This is particularly important when technological decisions are made at an organisational level. The benefits of automation may be visible in budgets, productivity measurements and performance indicators, while its consequences for employees, customers or citizens are more difficult to capture. What appears to be an improvement from one perspective may create new difficulties from another.
The challenge is not to reject efficiency, but to avoid confusing it with value. A more productive organisation is not necessarily one that creates better outcomes for the people it serves, just as a faster process is not necessarily one that strengthens the capabilities of those involved.
Before automating an activity, we therefore need to understand what makes that activity valuable and whether the proposed technology will preserve, diminish or transform that value.
What disappears when the task disappears?
Consider a young professional who uses artificial intelligence to prepare an analysis. Instead of spending several hours reading documents, comparing information and developing an argument, the professional can ask an AI system to produce a structured report within minutes. The result may be clear, comprehensive and sufficiently accurate for its intended purpose.
From a productivity perspective, the improvement is obvious. The same work can apparently be completed in a fraction of the time, allowing the professional to move on to other assignments. But this assessment assumes that producing the report was the only purpose of the activity.
In many professions, the process of preparing an analysis is also how expertise develops. Reading unfamiliar material, identifying contradictions, evaluating evidence and deciding which conclusions are justified are not merely obstacles on the way to a finished document. They are activities through which professionals learn to exercise judgement.
If artificial intelligence increasingly performs these activities, the professional may become more productive in the short term while having fewer opportunities to develop the capabilities required to assess the technology’s output. The ability to recognise an incomplete argument, identify a misleading conclusion or question an apparently authoritative recommendation depends on knowledge and experience that must be acquired somewhere.
A similar dilemma appears in education. A student may use AI to produce an excellent essay, complete with a coherent argument and well-structured language. If the purpose of the assignment is simply to produce a text, the technology may have performed remarkably well. But if the purpose is to develop the student’s ability to reason, organise ideas and communicate an argument, the quality of the finished essay tells us relatively little about whether learning has taken place.
This does not imply that students or professionals should avoid AI. Used differently, the same technology could support learning by helping people explore alternative arguments, understand difficult concepts or receive feedback on their reasoning. The important distinction lies in whether technology replaces the activity through which capabilities develop or supports people in developing those capabilities more effectively.
The question is therefore not simply which tasks can be delegated to machines. It is which human capabilities we want to preserve, strengthen or develop, and how technology can contribute to that ambition.
Sometimes removing a task creates an opportunity to become more capable. In other situations, removing the same task may eliminate an important opportunity to learn. Understanding the difference requires us to examine the purpose of the activity rather than focusing exclusively on the quality of its output.
Human involvement is not always the answer
Concerns about automation often lead to calls for keeping humans involved in technological processes. The idea of maintaining a human in the loop has become an important principle in discussions about artificial intelligence, particularly when automated systems influence decisions with significant consequences.
Human involvement can be valuable, but its presence does not automatically guarantee meaningful oversight or better outcomes. A person required to approve hundreds of algorithmic recommendations may have little opportunity to examine the reasoning behind each one. If the individual lacks sufficient information, time or authority to challenge the system, their involvement may provide the appearance of human control without its substance.
In such circumstances, keeping a person in the process can become a procedural formality rather than a meaningful safeguard. The technology continues to shape the decision, while the human participant is expected to accept responsibility for an outcome they may have had little realistic opportunity to influence.
The opposite problem also exists. Requiring people to perform repetitive or physically demanding activities simply because they have traditionally been carried out by humans does not necessarily protect human dignity or agency. For some individuals, automation is precisely what makes greater independence possible.
Speech recognition can enable people with limited hand function to write and communicate more easily. Automated translation and assistive communication technologies can help overcome barriers that would otherwise restrict participation. In the workplace, automation may remove administrative demands that prevent employees from concentrating on activities where their knowledge and judgement are most valuable.
These examples suggest that the relevant distinction is not between human and machine, but between different ways of organising their relationship. Technology can expand people’s ability to act independently, understand information and influence decisions. It can also reduce those opportunities by making individuals dependent on systems whose operation they cannot understand or challenge.
Meaningful human agency therefore cannot be measured simply by counting how many tasks remain under human control. It depends on whether people have genuine opportunities to understand their circumstances, make choices and influence what happens next.
An organisation might automate most of a process while strengthening human agency by giving people better information, more flexibility and greater control over important decisions. Another organisation might preserve human approval at every stage while leaving individuals with little real influence.
The design of the relationship matters more than the formal division of tasks.
Whose time are we saving?
Imagine a public institution introducing an automated system to handle citizen enquiries. The technology allows people to access information at any time, reduces waiting periods and decreases the number of enquiries requiring assistance from employees. From the institution’s perspective, the system may represent a substantial improvement in efficiency and service delivery.
For many citizens, it may be an improvement as well. Straightforward questions can be answered immediately, and routine procedures can be completed without waiting for an appointment or navigating a telephone queue. People who previously struggled to access services during conventional opening hours may gain greater flexibility and independence.
But the same system may create difficulties for people whose circumstances do not fit the expected pattern. Someone dealing with a complex personal situation may find it difficult to explain their needs through a standardised interface. A person with a cognitive disability may struggle to interpret the system’s instructions. Others may face language barriers or require assistance that cannot easily be reduced to a sequence of automated responses.
If the institution measures success primarily through reduced processing times and fewer human interactions, these experiences may be overlooked. A service that appears more efficient from the organisation’s perspective can become more demanding for some of the people who depend on it.
This illustrates a broader challenge in technological development. The costs and benefits of automation are not necessarily distributed equally. An organisation may save time by transferring responsibility to its customers. Employees may become more productive while losing opportunities to exercise professional judgement. Some users may gain independence, while others become more dependent on assistance from family members, support workers or alternative services.
Understanding these consequences requires more than technical testing and financial analysis. It requires perspectives from the people who will experience the technology in different circumstances.
Those perspectives can also reveal opportunities for better solutions. Perhaps the most valuable approach is not to automate every interaction, but to automate routine administration while making knowledgeable human assistance more accessible when it matters. Perhaps technology could reduce the burden of documentation and allow professionals to spend more time understanding individual circumstances. Perhaps the success of an automated service should be measured not only by how many interactions it eliminates, but also by how effectively it helps people resolve their problems.
Such possibilities are easy to overlook when the central objective is to reduce human involvement.
The more interesting question may be how automation can improve the quality of human interaction rather than simply reduce its frequency.
Automation changes more than work
Technologies do not merely replace individual activities. Over time, they can change the environments in which those activities take place and influence the capabilities people develop within them.
Navigation systems provide a familiar example. They make it easier to travel through unfamiliar places, reduce uncertainty and help people reach destinations efficiently. At the same time, relying heavily on automated navigation may change how people develop their understanding of geography and their ability to find their way independently.
Recommendation systems offer another example. They help people discover information, entertainment and products from an overwhelming range of possibilities. But by influencing which options are presented, they can also shape what people encounter and what they consider worth exploring.
Artificial intelligence extends these dynamics into activities that have traditionally depended on human reasoning and judgement. When AI systems generate analyses, recommend actions or propose solutions, they do more than save time. They may influence which alternatives people consider, which questions they ask and how they evaluate the information available to them.
These effects are not inherently positive or negative. Technologies have always changed human capabilities, and new tools can create forms of knowledge and expertise that did not previously exist. The widespread availability of calculators, for example, changed the relationship between mathematical work and manual calculation without eliminating the need for mathematical understanding.
What matters is that technological assistance can alter the skills and forms of judgement people practise. As certain activities become easier, others may become less familiar. New capabilities may emerge, while existing ones may weaken or become concentrated among fewer people.
Organisations can experience similar changes. Automating routine decisions may improve consistency and free employees for more complex work. But it may also reduce opportunities for junior professionals to gain experience through everyday practice. Over time, this could affect how expertise is developed and maintained within the organisation.
These changes are often gradual. They may not be visible in the immediate evaluation of a new technology, particularly when that evaluation focuses on productivity, accuracy or financial return. By the time their broader consequences become apparent, the technology may already be embedded in organisational routines and institutional practices.
This makes the decisions taken before automation especially important. They shape not only how work is performed today, but also which capabilities people and organisations may possess tomorrow.
A better basis for deciding what to automate
If automation changes more than the immediate task, decisions about its use should involve more than technical feasibility and financial return. Organisations need to understand the wider purpose of the activities they are considering automating and the consequences for the people involved.
The first question concerns the activity itself. What is it intended to achieve, and where does its value actually lie? Some tasks exist primarily to process information or move material from one place to another. Others contribute to learning, relationships, professional judgement or meaningful participation. These functions may coexist within the same activity, making it important to distinguish between elements that can be removed without significant loss and elements that deserve greater attention.
The second question concerns the opportunities automation creates. Could it increase independence, improve access to services or make new activities possible? Could it allow professionals to concentrate on more valuable work or enable organisations to deliver services that would otherwise be too expensive or difficult to provide? These possibilities should be considered alongside the immediate efficiency gains.
The third question concerns what may be lost. Automating a task could reduce opportunities to acquire expertise, weaken professional judgement or make people more dependent on technological systems. It may also change relationships between organisations and the people they serve. Such consequences do not automatically outweigh the benefits, but they deserve to be examined before decisions become difficult to reverse.
A fourth question concerns whose experiences are included in the assessment. A technology may produce different outcomes for managers, employees, customers and citizens. People with different abilities, levels of experience or access to resources may encounter opportunities and barriers that are invisible to those responsible for implementing the system. Understanding these differences can reveal both risks and possibilities that conventional evaluation methods overlook.
Finally, organisations should consider whether the choice is really between full automation and preserving existing practices. In many cases, the more valuable opportunity may lie in redesigning the activity itself. Technology might take responsibility for routine elements while people retain meaningful control over decisions. An automated system might support learning rather than replace the activities through which learning occurs. A service might become both more efficient and more personal if technology were used to remove administrative barriers rather than human interaction.
These considerations do not provide a universal formula for deciding what should be automated. Different circumstances will lead to different conclusions, and legitimate priorities may sometimes conflict. But they create a more useful basis for decision-making than the assumption that technological feasibility should determine the direction of change.
The objective should not be to preserve human activity regardless of its value, nor to automate every task that technology can perform. It should be to understand which combination of human and technological capabilities creates the greatest possibilities.
The more ambitious possibility
Automation is frequently presented as a way to do more with less. Organisations can produce more output with fewer resources, complete processes more quickly and reduce the amount of human effort required to perform familiar tasks. These improvements can be valuable, particularly when they free resources that can be used elsewhere.
But technological progress also offers an opportunity to reconsider what we are trying to achieve. Instead of asking only how existing activities can be performed more efficiently, we can ask whether technology makes different and more valuable activities possible.
In healthcare, automation might reduce administrative demands and create more time for care and personal interaction. In education, AI could support individual learning needs and give teachers greater opportunities to concentrate on understanding their students. In public services, automated systems could simplify routine procedures while making specialist assistance more accessible to people with complex circumstances.
In businesses, automation could enable employees to explore new ideas, develop better products or serve customers in ways that were previously impractical. For individuals, assistive technologies could expand independence and create opportunities for participation that existing systems have failed to provide.
None of these outcomes follows automatically from introducing technology. An organisation that reduces administrative work may choose to use the resulting savings to cut staffing rather than improve services. A system designed to support professional judgement may gradually become a mechanism for standardising decisions and restricting discretion. Technology creates possibilities, but organisational priorities and human decisions determine which of those possibilities are pursued.
That is why the future of automation cannot be understood simply as a question of technological capability. It is also a question of values, priorities and the kind of society we want to create.
We should be willing to automate activities that consume time without creating meaningful value. We should also be willing to preserve or redesign activities that contribute to learning, relationships, judgement and participation, even when technology could produce their immediate outputs more efficiently.
Most importantly, we should recognise that automation offers opportunities to create new forms of value rather than merely reduce the cost of existing ones.
The question is no longer whether machines can perform activities that once required human effort. Increasingly, they can. The more consequential question is what we choose to do with that capability.
The future of automation should not be determined by how much human activity we can eliminate, but by how much more human possibility we can create.
