IELTS Academic Reading Practice Test 158

When Wilhelm Röntgen discovered X-rays in 1895, he gave medicine its first way of seeing inside a living body. The new rays passed through soft tissue but were stopped by bone, and within months doctors were using them to locate fractures and bullets. The technique had an obvious weakness, however. An X-ray photograph is a flat shadow: every structure between the source and the film is superimposed on everything else, so a tumour inside the skull is hidden behind the dense bone that surrounds it. For nearly seventy years radiologists could do little about this except take pictures from several angles and try to imagine the three-dimensional structure that had produced them.

The man who solved the problem was not a doctor but an engineer. Godfrey Hounsfield was born in 1919 on a farm in Nottinghamshire and studied electrical engineering. During the Second World War he worked on radar with the Royal Air Force, and afterwards he joined EMI, a British company whose interests ranged from records to electronics. There he helped to build one of Britain's first all-transistor computers. That experience, unusual for an engineer of his generation, gave him the tool he would later need. It is often said that EMI's profits from its music business, including the Beatles, helped to pay for the research that followed.

Walking in the countryside in 1967, Hounsfield had an idea. If a narrow beam of X-rays were passed through the body from many different directions, and the amount absorbed were measured each time, a computer could work backwards from those measurements to calculate how much absorbing material lay in each small region of a slice through the body. A slice could be represented as a grid of thousands of tiny squares, or pixels, each with its own value. The mathematics was demanding and the computing power required was enormous by the standards of the day, but Hounsfield was convinced that the principle was sound.

Unknown to him, a South African-born physicist had already published the essential mathematics. Allan Cormack, working at Tufts University in Massachusetts, had become interested in reconstructing a cross-section from measurements taken along many lines, and in 1963 and 1964 he published papers showing how it could be done. He had even built a simple experimental apparatus, but his work attracted almost no attention. Hounsfield reached the same conclusion independently.

At EMI's laboratories in Hayes, west London, Hounsfield built the first machine. It was not a graceful object: a heavy frame, a rotating table and a rig that carried a source and a detector around the patient's head in a series of sweeps, taking readings at one degree intervals. Because the range of intensities passing through a head was too great for the detectors of the period to measure accurately, the head was surrounded by a bag of water, which evened out the signal. That water bag became a familiar feature of the first generation of scanners.

The first patient was examined on 1 October 1971 at Atkinson Morley's Hospital in Wimbledon, south London, where the neurosurgeon James Ambrose had agreed to test the machine. The patient had a suspected tumour in the frontal lobe of the brain. The scan took several minutes and produced, for the first time, a picture of the inside of a living human brain in which grey matter, white matter, fluid and bone could be distinguished. It showed the lesion clearly.

Ambrose's colleagues were impressed, and the results were presented at a congress in 1972. Orders followed from hospitals in Britain and the United States. The earliest machines could examine only the head, because the skull was small enough to be enclosed in the water bag and the reconstruction of a moving chest was beyond the computers then available; scanners for the whole body appeared later in the 1970s. By the end of the decade CT — originally 'computerised axial tomography', hence the older name CAT scan — had become an established part of hospital practice.

In 1979 Hounsfield and Cormack shared the Nobel Prize in Physiology or Medicine, an unusual honour for men working outside medicine, and Hounsfield was knighted two years later. The award recognised both the mathematics and the machine.

Modern scanners are different in almost every respect. The cables that once connected the rotating X-ray tube to the rest of the machine, and which limited it to one direction of travel, were replaced in the late 1980s by slip rings that allow continuous rotation. The table now moves steadily through the ring, so that the data describe a helix rather than a series of separate slices, and images can be reconstructed at any point along it. A modern scanner acquires hundreds of slices in the time a patient can hold a breath; a scan that once took hours to compute appears on a screen within seconds.

The main drawback is radiation. A CT examination delivers a dose many times greater than a plain X-ray — for a scan of the abdomen, roughly several years' worth of natural background radiation — and the risk rises with repeated scans, particularly in children. Radiologists therefore try to keep doses as low as reasonably practicable, and newer reconstruction methods can produce acceptable images from fewer measurements than were once thought necessary. Patients and doctors must weigh those risks against the information that a scan provides, and in many cases the balance favours having the examination.

A humanoid robot is one designed to resemble a person: a head, a torso, two arms and, in most cases, two legs. The category includes machines as different as Honda's ASIMO, which once greeted visitors at exhibitions, and Boston Dynamics' Atlas. Engineers give several reasons for copying the human frame. The strongest is that the world has been built for human bodies: door handles, staircases, ladders and vehicle pedals are all arranged for hands and feet of a particular size and reach. A robot with similar proportions can use existing tools and move through existing buildings without anyone having to alter them. A familiar shape may also make a machine less alarming to the people working beside it.

Not everyone is convinced. A robot that rolls on wheels is simpler, cheaper and far more stable, and in a factory or a hospital corridor there is little that a pair of legs can do and wheels cannot. Legs become essential only where the ground is uneven or where the machine has to climb, which is why legged robots have been pursued most energetically by companies interested in disaster response, construction sites and the delivery of parcels to people's doors. Even supporters of the humanoid form concede that the shape is usually chosen for compatibility with human environments rather than for any inherent superiority.

Walking on two legs is an exercise in controlled falling. The body behaves as an inverted pendulum, toppling forward while one leg swings ahead to catch it. A theoretical framework known as the zero moment point, published by the Serbian engineer Miomir Vukobratović in the late 1960s, described the conditions under which a machine would not tip over, and for three decades it underpinned almost every walking robot. Modern controllers instead calculate, many times a second, the forces that each joint must produce; a machine can then stay upright while running, jumping or being pushed.

The longest-running programme belonged to Honda. The company began work on a walking machine in 1986, when few outside research laboratories believed that a two-legged robot could be made practical. A series of prototypes followed, growing from a pair of legs that shuffled forward in a straight line to machines that could climb stairs and turn. The project was kept secret for years, and when a humanoid called ASIMO was shown to the public in 2000 it could walk, gesture and respond to spoken commands. Later versions ran and some worked as museum guides. Development nevertheless ended in 2018, when Honda chose to concentrate on other robotics work.

Today's most agile humanoids come from Boston Dynamics, a company founded in 1992 by a former MIT professor, Marc Raibert. Its Atlas robot, unveiled in 2013, was built for a competition sponsored by the American defence research agency DARPA, in which machines had to drive vehicles, clear rubble and climb stairs. Atlas was hydraulic: powerful, but noisy and inefficient. In 2024 the company replaced it with an electric version, whose joints move more smoothly and which is far quieter. Electric motors are also easier to control precisely, which matters when a machine has to place a foot on a narrow step.

A humanoid is a dense network of sensors. Every joint carries an encoder that reports its angle, an inertial measurement unit in the torso combines accelerometers and gyroscopes to track the body's orientation, and force sensors in the ankles and wrists measure how hard the machine pushes against the ground or against an object. All of this must be processed in milliseconds. Engineers increasingly rely on machine learning: a controller is trained inside a simulated copy of the robot, where it can fall thousands of times in an hour without damage, and the resulting program is then transferred to the real machine, a technique known as sim-to-real transfer.

The hands are, in many ways, harder than the legs. A human hand has more than twenty degrees of freedom, and its grip combines strength with delicacy: the same fingers can lift a suitcase and hold an egg. A robot hand with a separate motor for every joint would be heavy, costly and prone to breakdown, so engineers build underactuated hands, in which one actuator drives several fingers through a linkage and the fingers wrap themselves around whatever they grasp. Tactile sensors in the fingertips are also improving, allowing a machine to judge how firmly it is holding something before the object slips or breaks.

Energy remains the binding constraint. Walking on two legs is far less efficient than rolling, and a humanoid that must also power computers, sensors and arms may run for only an hour or two on a battery. Safety is another concern, because a machine with the mass and strength of a person can injure someone if it falls or miscalculates, and the components remain expensive. Designers must also consider psychology. In 1970 the roboticist Masahiro Mori suggested that people's response to a machine that appears almost human turns from affection to unease as it approaches likeness, a dip he called the uncanny valley.

Where are humanoids actually working? In warehouses a few machines are now employed alongside people, moving containers and stacking shelves; Agility Robotics' Digit is one example, and Amazon has tested such machines at its facilities. Others inspect power stations or handle laboratory samples. Many researchers believe that the next advance will come from 'foundation models' trained on large collections of human demonstration data, so that a robot can be shown a task instead of being programmed for it. Whether that will make the humanoid form commercially viable is still unclear.

The green dot beside a colleague's name is one of the small tyrannies of modern office life. It shows that the person is at their desk and available, and to leave a message unanswered while showing a green dot is, in many workplaces, a quiet statement of priorities. Messaging applications such as Slack, released in 2013, and Microsoft Teams, launched four years later, have spread through offices in little more than a decade, bringing a new set of questions about how people work.

The case for these tools is strong. Email, which they have partly displaced, was a poor instrument for conversation: long threads, copied recipients and no clear place for a discussion to live. A channel, by contrast, gathers messages about a single project in one searchable place, and a written record exists where once decisions were made in corridors and remembered by nobody. Because messages wait until the recipient is ready, teams spread across time zones can work together without meetings; a question asked in London at the end of the day can be answered in San Francisco by morning.

The case against is equally familiar. Employees describe being pulled between dozens of conversations, checking a phone before getting out of bed and feeling guilty when they switch off. It is tempting to blame the applications themselves. But the old world of email was never a haven of concentration: it, too, arrived continuously, and it, too, demanded replies. The difference is one of tempo. Messaging is designed for immediacy, and immediacy is what sustained thought cannot survive.

The psychology of interruption is well studied. Studies of office workers led by the psychologist Gloria Mark found that, once interrupted, people needed on average about 23 minutes to get back to the task they had left. The cost is not only the time lost. Each switch leaves what researchers call residue: part of the mind stays with the conversation that was abandoned, so the work resumed is done less well. A tool that invites a check every few minutes is therefore not a neutral convenience.

Messaging also makes work visible in a way that older tools did not. The green dot, the read receipt and the rapid reply are all signals that a person is present and busy. Managers under pressure to demonstrate that their teams are working may find such signals hard to resist, and staff quickly learn to produce them. Yet presence is not productivity. The colleague who answers instantly all day may simply be the one who has no uninterrupted hour in which to think.

This, I think, is where the real problem lies. It is not the software but the expectation, created by managers and accepted by everyone else, that a message will be answered at once. Once such a norm takes hold, it is rational for each person to keep checking, because the cost of appearing unresponsive falls on whoever waits while the benefit of concentration goes to the employer. The result is a collective problem that cannot be solved by individual willpower. Nor should staff be judged on how quickly they reply; they should be judged, as they always were, on what they produce.

What can be done? The most useful changes are organisational rather than technical. Teams can agree that messages will be answered in batches at set times rather than continuously, and that anyone who needs to concentrate may say so without apology. Notifications can be switched off by default, so that checking becomes a deliberate act rather than a reflex. Channels can be given clear purposes, with a few reserved for urgent matters and the rest allowed to wait.

Some countries have gone further. France requires larger employers to negotiate rules limiting the use of work devices outside working hours, and a right to disconnect has been debated or adopted in several others. Such laws are valuable, but they distribute their benefits unevenly. A salaried office worker in Paris can close a laptop at six; a hospital porter, a delivery driver whose route is set by an app, or a freelancer dependent on a single client cannot. Arguments about digital overload that ignore this difference are, in effect, arguments among the comfortable.

None of this suggests that the tools should be abandoned. The remedy for poor use of a technology is rarely to remove it; it is to decide collectively what it is for. Messaging applications are excellent at what they were designed to do, which is to let people ask quick questions and reach decisions. The productivity they promise will be realised only when organisations stop treating the speed of a reply as a measure of commitment. The green dot is not the problem. What we have decided it means is the problem.