IELTS CAM Inspired 21 - Reading Test 2 with Practice Test, Answers And Explanation
Luyện tập đề IELTS CAM Inspired 21 - Reading Test 2 được lấy từ cuốn sách IELTS CAM Inspired 21 với trải nghiệm thi IELTS trên máy và giải thích đáp án chi tiết bằng Linearthinking, kèm answer key và list từ vựng IELTS cần học trong bài đọc.
IELTS CAM Inspired 21 - Reading Test 2 with Practice Test, Answers And Explanation
Passage
❓ Câu hỏi passage 1
Question 1 - 5
Complete the table below.
Choose ONE WORD ONLY from the passage for each answer.
Research into sleep and dreaming
Subject
Findings
Comment
Humans
During REM sleep the muscles are held almost completely
1
so that the sleeper is not injured.
Two kinds of sleep alternate through the night.
Rats
The readings gathered while the rats moved across a
2
closely matched those from REM sleep.
Taken as a sign of dreaming.
Pigeons
While the birds slept, the most active areas were those processing the pigeon's vision, particularly pictures of travelling through the
3
.
Possibly dreaming about flight.
Whales and dolphins
Rather than putting the entire brain to sleep together, they keep one portion
4
.
They show no REM sleep.
Whatever dreams occur are likely to be relatively
5
.
Question 6 - 13
Do the following statements agree with the information given in the Reading Passage?
TRUE if the statement agrees with the information
FALSE if the statement contradicts the information
NOT GIVEN if there is no information on this
6
Studies of birds and mammals indicate that REM sleep does little to assist learning or memory.
7
Dreaming is thought to have become necessary for animals only after their ways of life grew moderately complex.
8
It cannot be established whether a dreaming dog has any inner experience of the actions it appears to perform.
9
Renner observed Marlowe in a tank located in his research laboratory.
10
Renner had kept octopuses in his home on earlier occasions.
11
As it slept, Bramble behaved as though it were under threat from a predator.
12
Renner regards observations of a single octopus as enough to prove that octopuses relive their waking experiences while dreaming.
13
Curiosity about the reasons behind human dreaming has grown markedly in recent years.
Passage
❓ Câu hỏi passage 2
Question 14 - 19
Reading Passage 2 has seven paragraphs, A-G.
Which paragraph contains the following information?
Write the correct letter, A-G.
14
a figure for how large the population of Mapungubwe eventually became
15
a reference to crops and livestock being traded for other necessary goods
16
a description of traders who reached the kingdom from across the sea
17
the place where the ruler's male kinsmen are thought to have lived
18
an admission that the role played by some items is not understood
19
the suggestion that a spell of harsh weather worsened the kingdom's difficulties
Question 20 - 21
Choose TWO letters, A-E.
The passage gives several details about gold in the kingdom of Mapungubwe. Which TWO of the following does the writer mention?
A
Gold circulated within Mapungubwe as a kind of money.
B
No item found at the site was more precious than a gold sceptre.
C
Certain gold pieces lay close to where the kingdom's rulers were buried.
D
Fashioning objects from gold was regarded as a high-status craft.
E
A way of decorating gold was also practised in a second kingdom.
Question 22 - 26
Complete the summary below.
Choose ONE WORD ONLY from the passage for each answer.
Archaeological discoveries
The large quantity of skilfully produced pottery found at the site points to the of the Mapungubwe community. Many of these items, among them beakers and bowls, are richly decorated with stamped patterns. The site has also yielded flat circular pieces of pottery, and models of several kinds of animal, together with figures of people whose were made unnaturally long. These figures may have played a part in rituals held to honour . Also uncovered were small made from a metal obtained in the local area.
Passage
❓ Câu hỏi passage 3
Question 27 - 29
Choose the correct letter, A, B, C or D.
Write the correct letter in boxes on your answer sheet.
27
In the first paragraph, the writer is
A
characterising the way the public regards AI.
B
forecasting the effect AI will have in years to come.
C
listing several concrete advantages that AI offers.
D
stressing how widely AI already reaches into daily life.
28
The writer's chief purpose in the second paragraph is to
A
identify the most enthusiastic supporters of AI solutionism.
B
trace where the idea of AI solutionism most likely came from.
C
draw attention to a danger that AI solutionism carries.
D
weigh the benefits of AI solutionism against its drawbacks.
29
What does the writer say about many politicians in the fourth paragraph?
A
They fail to appreciate how difficult putting large-scale AI schemes into practice really is.
B
They have misunderstood what the machine-learning sector is for.
C
They have overlooked the genuine potential of such schemes.
D
They are too eager to join the contest to lead machine learning.
Question 30 - 35
Complete the summary using the list of words, A-I, below.
Write the correct letter, A-I, in boxes on your answer sheet.
AI in government, medicine and the law
Neural networks are among the more encouraging branches of AI for governments. Yet many politicians overrate what they can do, assuming that simply attaching a neural network to a problem will produce answers and, at a stroke, promote greater in the way citizens are dealt with.
Most public bodies, however, have never established the required to cope with the enormous quantities of data that AI depends on. Bureaucracy makes matters worse: because information is scattered across separate departments, every official involved must obtain before the relevant data can be reached. The deeper problem is that hardly any of their staff possess the needed to exploit machine intelligence fully.
In medicine, an AI tool was tried, but the experts who used it had little faith in the of its guidance, and it was dropped. The courts likewise stopped using algorithms once it emerged that these intensified the already embedded in the country's legal system.
A
authorisation
B
bias
C
dependability
D
expertise
E
fairness
F
groundwork
G
hardware
H
investment
I
secrecy
Question 36 - 39
Do the following statements agree with the information given in Reading Passage?
In following statements below, choose
YES if the statement agrees with the information
NO if the statement contradicts the information
NOT GIVEN if it is impossible to say what the writer thinks about this
36
Voss believes that attention should be directed towards the ordinary, practical uses of AI.
37
Nandakumar claims that advances in robotics and AI reach widespread use more quickly than experts expect.
38
Defending AI systems against attack costs more than most organisations anticipate.
39
The writer argues that public mistrust of AI can be safely disregarded once the technology works well.
Question 40
Choose the correct letter, A, B, C or D.
Write the correct letter in boxes on your answer sheet.
40
Which of the following would be the most suitable subtitle for Reading Passage 3?
A
Making the most of everything AI has to offer
B
Why governments ought not to fund AI systems
C
How AI could raise the efficiency of public services
D
Why AI is not the answer to every problem
DOL IELTS Đình Lực
21/10/2026
📖 Bài đọc passage 1
Do animals dream?
When electrodes are attached to a sleeper's head and monitored overnight, two separate kinds of sleep emerge, and the body alternates between them all night long. One is active, or rapid eye movement (REM), sleep, so named because the eyes keep shifting beneath the closed lids; the muscles give occasional faint twitches, yet are largely held immobilised, which keeps the sleeper from coming to any harm. The other, non-REM sleep, is a much stiller state with hardly any movement. Most of our dreaming occurs in REM sleep, and the most lifelike dreams of all belong to it.
Other mammals appear to dream too. In one study, the brain activity of rats picking their way through a maze was set beside the activity recorded from the same animals in REM sleep; the readings matched so closely that researchers concluded the sleeping rats were re-living their run through the maze. Showing the same in non-mammals has been far harder, both because their brains are so unlike ours and because monitoring a sleeping brain is awkward. Even so, one team lately recorded the sleeping brains of pigeons, and, as in mammals, both REM and non-REM phases appeared. Strikingly, during REM sleep the busiest regions were those handling what the bird sees, especially images linked to moving through the air; if the pigeons were dreaming, flight seems a likely subject.
None of this means dreaming, or REM sleep, occurs throughout the animal world. Sponges are an obvious exception: with no brain, they lack any apparatus a dream would need. Some animals also sleep in unusual ways. Whales and dolphins, for instance, never shut the whole brain down at once but put only one half to sleep at a time, keeping the other half alert. They show no REM sleep either, which hints that any dreams they have are of the non-REM sort, and so the fainter kind — surprising, given the rich inner lives we credit them with. The usual explanation is that REM sleep is skipped because it would leave them more exposed to extremes of temperature.
Even so, REM sleep does seem to pay off in many cases. A growing body of research on warm-blooded animals suggests that REM sleep and dreaming matter for learning and for laying down memories. The thinking is that replaying an event in a dream helps fix that memory into longer-term storage. On this view, once animals adopted even moderately complicated ways of living, dreaming became a necessity for managing them.
What we still cannot explain is how this visible behaviour relates to whatever the animal inwardly feels. Getting inside the mind of a rat or a pigeon seems impossible, let alone picturing the scenery of their dreams. When a sleeping dog's legs twitch and it gives soft, muffled barks, we rush to interpret the scene; in truth, we have no way of knowing whether any inner sense of chasing a rabbit accompanies what we see.
A very different non-human dreamer is instructive here. In 2020, while filming a wildlife documentary, the marine biologist Tobias Renner, of the Halvorsen Institute of Marine Biology, gave a home to an octopus he named Marlowe, keeping her in a tank in his own living room. One night Marlowe seemed to dream: her arms and head stirred and her skin raced through rapid colour changes, much as if she were closing in on a crab.
A comparable account appeared not long ago, this time of a sleeping octopus seemingly caught in a nightmare. Named Bramble, the creature thrashed about, pushed out his mantle as if swelling to a more imposing size, and shot out ink, exactly as he would if a predator were bearing down on him. Renner finds the nightmare account intriguing but notes that it rests on just one animal. Watching an octopus's behaviour, he argues, cannot settle the matter alone; only brain imaging could show that these animals are working back through episodes of their waking lives as they dream.
The real difficulty is that any animal's dreams will always lie beyond our direct reach — as, for that matter, do other people's. Still, we can try to imagine those inner worlds by taking each animal on its own terms. Most humans, for instance, depend on sight above every other sense, which is why the visual dominates our dreams. Dogs, by contrast, find their way largely by scent, while spiders rely far more on the vibrations they detect.
In all likelihood, dreaming has served several distinct purposes ever since the first complex animals arose. If so, a fuller grasp of those purposes might, in turn, illuminate what our own dreaming is really for.
Rising on an inland plateau beside the upper reaches of the Nzembe River, the kingdom of Mapungubwe rose to prominence between the 11th and 14th centuries CE.
A
A. Among the earliest states to emerge in the region, Mapungubwe was established by Tsivara-speaking peoples who lived chiefly by farming. At the core of the territory its rulers governed lay a broad sandstone plateau whose sheer inaccessibility made it simple to defend. Food was plentiful here, as in neighbouring kingdoms, for herding cattle and growing crops yielded not only enough to eat but a surplus that could be exchanged for whatever the people lacked. Thick deposits of bone and dung point to sizeable cattle herds from the 8th century CE onward, and in this part of the world large herds had long been the customary basis of a ruler's power. By the 9th century domesticated cattle had grown markedly more numerous, while plentiful spindle whorls reveal that cotton was by then being grown and woven.
B
At its height, most likely in the early 13th century, Mapungubwe may have been home to some 4,000 inhabitants. Its chief, whom people called the king, is thought to have been the richest of all, holding more cattle and valuables than any of his subjects. He and his counsellors occupied a walled compound of stone dwellings set on the very highest ground the kingdom possessed: a natural sandstone hill rising roughly 25 metres and stretching about 120 metres from one end to the other. People had lived on this hill since the 10th century, and postholes cut into the rock reveal that a wooden palisade once ringed the entire compound. Everyone else was housed in mud and thatch dwellings on the flat land below, where a single stone building also stood. That lower settlement, called Tshiluvi, spreads across some 6 hectares (14.8 acres) and is older than anything raised on the hilltop.
C
Whereas ordinary people were laid to rest down in the valley encircling the site, the kings themselves were buried apart, within a marked-off zone at the hill's summit. A wooden staircase joined the two burial levels, and the holes that once held its steps can still be made out in the sandstone cliff face. Scattered around the edge of Tshiluvi stood a number of unusually large houses, generally taken to have been the homes of the king's male kin. Across the plateau lie many further sites, more modest than the central hill yet impressive in their own right, some as near as 20 and others as far as 90 kilometres away. Built of stone walls and dwellings, these most probably belonged to minor chiefs who served the king.
D
Bones of large predators and slivers of ivory turn up in great quantity on the plateau, a hint that the hides of such beasts, along with elephant tusks, were being stockpiled, probably to be carried down the Nzembe to the coast. Glass beads found at the site, which can only have come from the ports of Adhrima, show that the kingdom traded in some fashion with coastal states, and those states in turn dealt with merchants who arrived by sea. Copper from nearby sources added to its wealth, as did the gold that travelled down from Marovanya, a kingdom to the north (itself prominent from the 12th to the 15th century), on its way to the coastal city of Selora. Command over such profitable long-distance dealings very likely tightened the grip of Mapungubwe's rulers, who had every reason to control, and if possible monopolise, them.
E
The sheer quantity of expertly made pottery unearthed at the site is taken as further proof of the wealth Mapungubwe society had attained, since such output implies that potting had become a full-time profession. Among the finds are round-bodied, short-necked jars, along with beakers and bowls, a great many stamped with decorative motifs. Ceramic discs and rattles have likewise come to light, as have figurines of cattle, sheep and goats and small, highly stylised human figures given elongated necks and stunted arms. What part these human figures played remains uncertain, though they may have served as offerings to the gods during ceremonies. Small ornaments worked from copper of local origin round out the discoveries.
F
Finely worked objects of gold have likewise been recovered from Mapungubwe. One decorative technique, unknown anywhere else but at Marovanya, called for gold to be beaten into little rectangular plates into which geometric designs were then incised. Using tiny gold tacks, craftsmen fixed these plates over objects of wood, none of which has lasted to the present day. A sceptre may be among the items treated in this way; further signs of gold-working on the spot include a rhinoceros modelled from small beaten plates and many thousands of tiny gold beads. Every one of these pieces came from the royal graves and belongs to about the year 1210. They stand as the earliest evidence anywhere in the region that gold was prized in its own right, rather than merely as a form of money.
G
By the opening years of the 14th century Mapungubwe had already begun to fade. The likeliest cause was that too large a population had come to press too heavily on local resources, a strain a run of droughts may well have driven to breaking point. The trade routes, too, may have shifted northward; at any rate it was the northern territories, Marovanya chief among them, that now prospered.
Across much of the wealthy world, anxious speculation about where artificial intelligence (AI) is heading has become impossible to escape. Barely a day passes without some breathless report claiming that AI will wipe out illness, speed up scientific discovery or make people more inventive. Judging by the headlines alone, one might conclude that AI has already reached into every corner of our lives.
That AI has created many genuine openings is not in doubt. Alongside them, however, a way of thinking has taken hold that might be labelled 'AI solutionism' - the conviction that machine-learning algorithms, if fed enough data, can resolve every problem humanity faces. Far from advancing the technology, this outlook undermines the worth of machine intelligence, since it brushes aside essential principles of AI safety and encourages hopes about the technology that it cannot possibly meet.
It has taken only a handful of years for this belief to travel from the sales pitches of industry enthusiasts in the technology sector to officials and policymakers on almost every continent. Where once the fashionable fear was that AI would ruin humanity, the mood has swung to the opposite extreme - the cheerful assumption that a computational rescuer has finally appeared.
One government after another is now promising money for national AI programmes and jostling for position in a race to lead the fast-expanding machine-learning industry. Yet for all the talk among politicians of a world-changing 'AI revolution' just around the corner, what tends to elude them is how complicated it is to put sophisticated machine-learning systems to work outside the laboratory.
Among the more encouraging kinds of AI are neural networks, a style of machine learning that borrows, in simplified form, from the way neurons are arranged in the human brain. What escapes many politicians, though, is that bolting a neural network onto a problem is no guarantee that a solution will appear. By the same logic, feeding one into the machinery of government will not, at a stroke, render it any more inclusive or even-handed.
Machine-learning tools are hungry for data, and here the public sector is poorly equipped: the digital infrastructure that advanced systems demand is, in most cases, simply not there. A great deal of official information still sits in offline records, and what little has been digitised is frequently lost inside layers of bureaucracy. Because such data is split between departments, reaching any part of it calls for separate permissions each time. Worst of all, government bodies rarely employ people with the technical know-how needed to draw real value from machine intelligence.
Unsurprisingly, the hype has drawn its share of sceptics. Aleksander Voss, who teaches computer science at Halvorsen University, has for years pressed for a cooler, more grounded outlook - one that attends to the plain, day-to-day uses of AI rather than to imagined scenarios in which super-intelligent machines seize control. A comparable note is struck by Priya Nandakumar, a robotics specialist at the Fenwick Institute of Technology, who observes that 'nearly every advance in robotics and AI takes far, far longer to come into general use than those inside the field, and those outside it, tend to assume'.
A further obstacle to putting machine-learning systems into service is their marked vulnerability to adversarial attacks, in which one hostile AI manipulates another into behaving as intended - pushing it, for instance, towards mistaken predictions. Although several researchers have cautioned that AI should not be released without proper security standards and defensive measures in place, the protection of these systems is still, all too often, treated as an afterthought.
If the gains from AI are to be realised and its dangers held in check, the conversation has to shift towards where, precisely, machine learning can be put to worthwhile use within government, commerce and society. That, in turn, means confronting questions of AI ethics head-on, along with the deep suspicion that many people continue to feel towards the technology.
Above all, we would do well to recognise where AI falls short and where human judgement must still lead the way. Rather than talking up its powers, the wiser course is to pause and draw a line between what the technology can actually do and what belongs to wishful thinking.
Medicine, too, has run up against AI's limits. A system known as OncoAssist was built to support doctors in treating cancer, and was designed to put forward the strongest possible course of action in each case. Even so, the specialists who used it struggled to place their confidence in its recommendations, and before long it had been dropped by most of the hospitals that gave it a trial.
Something similar unfolded in the courts, where sentencing software was brought in to help decide the fate of convicted offenders. The programme worked out risk scores and passed recommendations to judges, but it was eventually shown to deepen the very patterns of discrimination that already ran through the justice system, and its use was discontinued.
For anyone hoping to pour more money into national AI programmes, there are important lessons here. What these cases make plain is that no single AI fix exists for every situation. Reaching for AI merely because it is available will not always prove worthwhile, and not every problem is best tackled by turning machine intelligence loose on it. Every solution carries a price, and the fact that a task can be automated does not mean it ought to be.
When electrodes are attached to a sleeper's head and monitored overnight, two separate kinds of sleep emerge, and the body alternates between them all night long. One is active, or rapid eye movement (REM), sleep, so named because the eyes keep shifting beneath the closed lids; the muscles give occasional faint twitches, yet are largely held immobilised, which keeps the sleeper from coming to any harm. The other, non-REM sleep, is a much stiller state with hardly any movement. Most of our dreaming occurs in REM sleep, and the most lifelike dreams of all belong to it.
Other mammals appear to dream too. In one study, the brain activity of rats picking their way through a maze was set beside the activity recorded from the same animals in REM sleep; the readings matched so closely that researchers concluded the sleeping rats were re-living their run through the maze. Showing the same in non-mammals has been far harder, both because their brains are so unlike ours and because monitoring a sleeping brain is awkward. Even so, one team lately recorded the sleeping brains of pigeons, and, as in mammals, both REM and non-REM phases appeared. Strikingly, during REM sleep the busiest regions were those handling what the bird sees, especially images linked to moving through the air; if the pigeons were dreaming, flight seems a likely subject.
None of this means dreaming, or REM sleep, occurs throughout the animal world. Sponges are an obvious exception: with no brain, they lack any apparatus a dream would need. Some animals also sleep in unusual ways. Whales and dolphins, for instance, never shut the whole brain down at once but put only one half to sleep at a time, keeping the other half alert. They show no REM sleep either, which hints that any dreams they have are of the non-REM sort, and so the fainter kind — surprising, given the rich inner lives we credit them with. The usual explanation is that REM sleep is skipped because it would leave them more exposed to extremes of temperature.
Even so, REM sleep does seem to pay off in many cases. A growing body of research on warm-blooded animals suggests that REM sleep and dreaming matter for learning and for laying down memories. The thinking is that replaying an event in a dream helps fix that memory into longer-term storage. On this view, once animals adopted even moderately complicated ways of living, dreaming became a necessity for managing them.
What we still cannot explain is how this visible behaviour relates to whatever the animal inwardly feels. Getting inside the mind of a rat or a pigeon seems impossible, let alone picturing the scenery of their dreams. When a sleeping dog's legs twitch and it gives soft, muffled barks, we rush to interpret the scene; in truth, we have no way of knowing whether any inner sense of chasing a rabbit accompanies what we see.
A very different non-human dreamer is instructive here. In 2020, while filming a wildlife documentary, the marine biologist Tobias Renner, of the Halvorsen Institute of Marine Biology, gave a home to an octopus he named Marlowe, keeping her in a tank in his own living room. One night Marlowe seemed to dream: her arms and head stirred and her skin raced through rapid colour changes, much as if she were closing in on a crab.
A comparable account appeared not long ago, this time of a sleeping octopus seemingly caught in a nightmare. Named Bramble, the creature thrashed about, pushed out his mantle as if swelling to a more imposing size, and shot out ink, exactly as he would if a predator were bearing down on him. Renner finds the nightmare account intriguing but notes that it rests on just one animal. Watching an octopus's behaviour, he argues, cannot settle the matter alone; only brain imaging could show that these animals are working back through episodes of their waking lives as they dream.
The real difficulty is that any animal's dreams will always lie beyond our direct reach — as, for that matter, do other people's. Still, we can try to imagine those inner worlds by taking each animal on its own terms. Most humans, for instance, depend on sight above every other sense, which is why the visual dominates our dreams. Dogs, by contrast, find their way largely by scent, while spiders rely far more on the vibrations they detect.
In all likelihood, dreaming has served several distinct purposes ever since the first complex animals arose. If so, a fuller grasp of those purposes might, in turn, illuminate what our own dreaming is really for.
B. At its height, most likely in the early 13th century, Mapungubwe may have been home to some 4,000 inhabitants. Its chief, whom people called the king, is thought to have been the richest of all, holding more cattle and valuables than any of his subjects. He and his counsellors occupied a walled compound of stone dwellings set on the very highest ground the kingdom possessed: a natural sandstone hill rising roughly 25 metres and stretching about 120 metres from one end to the other. People had lived on this hill since the 10th century, and postholes cut into the rock reveal that a wooden palisade once ringed the entire compound. Everyone else was housed in mud and thatch dwellings on the flat land below, where a single stone building also stood. That lower settlement, called Tshiluvi, spreads across some 6 hectares (14.8 acres) and is older than anything raised on the hilltop.
C. Whereas ordinary people were laid to rest down in the valley encircling the site, the kings themselves were buried apart, within a marked-off zone at the hill's summit. A wooden staircase joined the two burial levels, and the holes that once held its steps can still be made out in the sandstone cliff face. Scattered around the edge of Tshiluvi stood a number of unusually large houses, generally taken to have been the homes of the king's male kin. Across the plateau lie many further sites, more modest than the central hill yet impressive in their own right, some as near as 20 and others as far as 90 kilometres away. Built of stone walls and dwellings, these most probably belonged to minor chiefs who served the king.
D. Bones of large predators and slivers of ivory turn up in great quantity on the plateau, a hint that the hides of such beasts, along with elephant tusks, were being stockpiled, probably to be carried down the Nzembe to the coast. Glass beads found at the site, which can only have come from the ports of Adhrima, show that the kingdom traded in some fashion with coastal states, and those states in turn dealt with merchants who arrived by sea. Copper from nearby sources added to its wealth, as did the gold that travelled down from Marovanya, a kingdom to the north (itself prominent from the 12th to the 15th century), on its way to the coastal city of Selora. Command over such profitable long-distance dealings very likely tightened the grip of Mapungubwe's rulers, who had every reason to control, and if possible monopolise, them.
E. The sheer quantity of expertly made pottery unearthed at the site is taken as further proof of the wealth Mapungubwe society had attained, since such output implies that potting had become a full-time profession. Among the finds are round-bodied, short-necked jars, along with beakers and bowls, a great many stamped with decorative motifs. Ceramic discs and rattles have likewise come to light, as have figurines of cattle, sheep and goats and small, highly stylised human figures given elongated necks and stunted arms. What part these human figures played remains uncertain, though they may have served as offerings to the gods during ceremonies. Small ornaments worked from copper of local origin round out the discoveries.
F. Finely worked objects of gold have likewise been recovered from Mapungubwe. One decorative technique, unknown anywhere else but at Marovanya, called for gold to be beaten into little rectangular plates into which geometric designs were then incised. Using tiny gold tacks, craftsmen fixed these plates over objects of wood, none of which has lasted to the present day. A sceptre may be among the items treated in this way; further signs of gold-working on the spot include a rhinoceros modelled from small beaten plates and many thousands of tiny gold beads. Every one of these pieces came from the royal graves and belongs to about the year 1210. They stand as the earliest evidence anywhere in the region that gold was prized in its own right, rather than merely as a form of money.
G. By the opening years of the 14th century Mapungubwe had already begun to fade. The likeliest cause was that too large a population had come to press too heavily on local resources, a strain a run of droughts may well have driven to breaking point. The trade routes, too, may have shifted northward; at any rate it was the northern territories, Marovanya chief among them, that now prospered.
Across much of the wealthy world, anxious speculation about where artificial intelligence (AI) is heading has become impossible to escape. Barely a day passes without some breathless report claiming that AI will wipe out illness, speed up scientific discovery or make people more inventive. Judging by the headlines alone, one might conclude that AI has already reached into every corner of our lives.
That AI has created many genuine openings is not in doubt. Alongside them, however, a way of thinking has taken hold that might be labelled 'AI solutionism' - the conviction that machine-learning algorithms, if fed enough data, can resolve every problem humanity faces. Far from advancing the technology, this outlook undermines the worth of machine intelligence, since it brushes aside essential principles of AI safety and encourages hopes about the technology that it cannot possibly meet.
It has taken only a handful of years for this belief to travel from the sales pitches of industry enthusiasts in the technology sector to officials and policymakers on almost every continent. Where once the fashionable fear was that AI would ruin humanity, the mood has swung to the opposite extreme - the cheerful assumption that a computational rescuer has finally appeared.
One government after another is now promising money for national AI programmes and jostling for position in a race to lead the fast-expanding machine-learning industry. Yet for all the talk among politicians of a world-changing 'AI revolution' just around the corner, what tends to elude them is how complicated it is to put sophisticated machine-learning systems to work outside the laboratory.
Among the more encouraging kinds of AI are neural networks, a style of machine learning that borrows, in simplified form, from the way neurons are arranged in the human brain. What escapes many politicians, though, is that bolting a neural network onto a problem is no guarantee that a solution will appear. By the same logic, feeding one into the machinery of government will not, at a stroke, render it any more inclusive or even-handed.
Machine-learning tools are hungry for data, and here the public sector is poorly equipped: the digital infrastructure that advanced systems demand is, in most cases, simply not there. A great deal of official information still sits in offline records, and what little has been digitised is frequently lost inside layers of bureaucracy. Because such data is split between departments, reaching any part of it calls for separate permissions each time. Worst of all, government bodies rarely employ people with the technical know-how needed to draw real value from machine intelligence.
Unsurprisingly, the hype has drawn its share of sceptics. Aleksander Voss, who teaches computer science at Halvorsen University, has for years pressed for a cooler, more grounded outlook - one that attends to the plain, day-to-day uses of AI rather than to imagined scenarios in which super-intelligent machines seize control. A comparable note is struck by Priya Nandakumar, a robotics specialist at the Fenwick Institute of Technology, who observes that 'nearly every advance in robotics and AI takes far, far longer to come into general use than those inside the field, and those outside it, tend to assume'.
A further obstacle to putting machine-learning systems into service is their marked vulnerability to adversarial attacks, in which one hostile AI manipulates another into behaving as intended - pushing it, for instance, towards mistaken predictions. Although several researchers have cautioned that AI should not be released without proper security standards and defensive measures in place, the protection of these systems is still, all too often, treated as an afterthought.
If the gains from AI are to be realised and its dangers held in check, the conversation has to shift towards where, precisely, machine learning can be put to worthwhile use within government, commerce and society. That, in turn, means confronting questions of AI ethics head-on, along with the deep suspicion that many people continue to feel towards the technology.
Above all, we would do well to recognise where AI falls short and where human judgement must still lead the way. Rather than talking up its powers, the wiser course is to pause and draw a line between what the technology can actually do and what belongs to wishful thinking.
Medicine, too, has run up against AI's limits. A system known as OncoAssist was built to support doctors in treating cancer, and was designed to put forward the strongest possible course of action in each case. Even so, the specialists who used it struggled to place their confidence in its recommendations, and before long it had been dropped by most of the hospitals that gave it a trial.
Something similar unfolded in the courts, where sentencing software was brought in to help decide the fate of convicted offenders. The programme worked out risk scores and passed recommendations to judges, but it was eventually shown to deepen the very patterns of discrimination that already ran through the justice system, and its use was discontinued.
For anyone hoping to pour more money into national AI programmes, there are important lessons here. What these cases make plain is that no single AI fix exists for every situation. Reaching for AI merely because it is available will not always prove worthwhile, and not every problem is best tackled by turning machine intelligence loose on it. Every solution carries a price, and the fact that a task can be automated does not mean it ought to be.
