雅思阅读 92: The Machine That Decides Who Dies（决定生死的机器）

改编自 Stop Killer Robots / United Nations（2026年8月）。雅思阅读 Section 3 难度，约 1050 词。 素材来源：https://egypt.un.org/en/321747-un-chief-red-cross-renew-call-rules-lethal-autonomous-weapons

Reading Passage

A. For as long as weapons have existed, a human being has had to pull the trigger, release the bowstring or launch the missile. That is now changing. Modern military systems can already identify a target, track it, decide whether it is hostile, and fire — all in a fraction of a second, far faster than a human supervisor could intervene. What is not yet fielded at scale is a system in which the choice of who to kill is made, end to end, by software alone. Weapons designers call this "lethal autonomous functionality"; campaigners call the resulting systems "killer robots". The ethical and legal arguments about whether such weapons should be allowed have been running for more than a decade, and in August 2026 the Secretary-General of the United Nations and the President of the International Committee of the Red Cross issued a joint appeal demanding that governments finally act. Innovation, they wrote, must serve humanity, not endanger it. The technology, meanwhile, continues to improve. The range of sensors, drones and cheap autonomous vehicles available off the civilian market has made the barrier to deployment lower every year, while the ethical debate has moved at the speed of intergovernmental committees. Civilian protesters, scientists and a growing number of governments want the same outcome: an international treaty that draws a line before the line is crossed.

B. The definition that matters is narrower than the phrase "killer robots" suggests. Not every military use of artificial intelligence is controversial. A drone whose human operator, sitting thousands of kilometres away, watches a feed and presses a button is not autonomous. A missile that locks on to an aircraft after the launch order has been given is at the edge. What campaigners want to prohibit is a system that, once activated, can itself detect a target, classify it as hostile and apply lethal force to it, with no further human approval — and, crucially, can do so in unpredictable ways. A two-tier regulatory regime has emerged through decades of negotiation. The first tier is a flat prohibition on autonomous weapons designed to target human beings directly, the so-called anti-personnel systems. The second tier regulates, rather than bans, autonomous weapons used against military objects, requiring predictability, explainability and meaningful human control over their use.

C. The moral case for the prohibition rests on three overlapping claims. The first is dignity: a machine, however sophisticated, does not understand that it is ending a human life. It does not grieve, does not hesitate, and cannot grasp the difference between a combatant who is surrendering and one who is fighting. A life-and-death decision made by an algorithm reduces the person on the receiving end to a data vector, a pattern in a sensor feed. The second is accountability. If an autonomous system kills a civilian in error, who is responsible — the programmer, the commander, the manufacturer, the officer who switched it on? Existing law, built around the chain of command, struggles to answer. The third is unpredictability. Machine-learning systems behave differently when confronted with situations their designers never imagined. A weapons system that cannot be expected to behave as its operators intend is, in effect, an uncontrolled projectile.

D. Proliferation is the fourth, and perhaps most persuasive, argument. Once the technology exists and works reliably, it will be cheap to copy and hard to contain. Unlike nuclear weapons, which require enriched uranium and large industrial plants, autonomous drones and ground robots can be built from off-the-shelf sensors, commercial chips and open-source algorithms. Small states, non-state armed groups and even criminal organisations could plausibly field them. A world in which border patrols, checkpoints and urban warfare are conducted by machines that can be ordered to engage without further human oversight is a world in which the barrier to killing drops almost to zero. Campaigners draw a deliberate parallel with the treaties that banned chemical weapons and anti-personnel landmines: those instruments worked because governments agreed, before the weapons became ubiquitous, that some weapons ought not to exist.

E. Governments have not, on the whole, disagreed with the principle. What they have refused is a treaty. Negotiations in the Convention on Certain Conventional Weapons, a forum at the United Nations, have run for years without agreement, in part because the major military powers want to preserve freedom to develop what they call "autonomy in support of the warfighter" — defensive systems that intercept incoming missiles without human reaction time. The 2026 joint appeal from the UN and the Red Cross is therefore framed as a last deadline, a reminder that delay itself is a choice. No country has yet publicly fielded a fully autonomous weapon that selects and engages human targets, but prototypes exist, and in at least one recent conflict, United Nations investigators have documented systems that operated at the very edge of meaningful human control. Diplomatic negotiations, which once had the luxury of decades, are now racing against a technology curve measured in months. The question for the next decade is not whether the technology will mature, but whether human societies will decide, before it does, that some decisions — including the one that ends a life — should remain, in a phrase that has become the movement's motto, meaningfully human. Campaigners often note that the strongest argument for keeping a human in the loop is not abstract morality but prudence: machines make mistakes, those mistakes kill, and the trail of error is easier to follow when someone was watching. If the past decade of autonomous-car crashes is any guide, even the most advanced models fail in corner cases that no training set prepared them for. Weapons, by contrast, do not have the luxury of pulling over to the side of the road when the algorithm is unsure.

Questions 1-4

Choose the correct heading for paragraphs B, C, D and E from the list of headings below.

List of Headings i. What is and is not controversial autonomy ii. The moral case for prohibition iii. Proliferation and the landmine parallel iv. Why governments have not yet agreed on a treaty v. The history of chemical weapons treaties vi. How drones are piloted from thousands of kilometres away vii. The design of open-source military software

1. Paragraph B: ____
2. Paragraph C: ____
3. Paragraph D: ____
4. Paragraph E: ____

Questions 5-8

Choose the correct letter, A, B, C or D.

5. What, according to the passage, is the central feature of a truly "autonomous" weapon? A. It is piloted remotely by a human operator. B. It can detect, classify and engage a target without further human approval. C. It uses artificial intelligence to help a human operator aim. D. It is made entirely from commercial parts.

6. Which of the following is NOT one of the three moral arguments given in paragraph C? A. Machines cannot understand human dignity. B. Accountability for errors becomes unclear. C. Such weapons are more expensive than conventional ones. D. Machine-learning behaviour is unpredictable.

7. Why do campaigners compare autonomous weapons to landmines? A. Both are used only at sea. B. Both should be banned before they become widespread. C. Both require enriched uranium. D. Both were invented by the same country.

8. Why have major military powers resisted a treaty? A. They want to keep the option of developing defensive autonomous systems. B. They believe autonomous weapons are unethical. C. They have already stopped all autonomous weapons research. D. They argue that open-source software should be free.

Questions 9-13

Do the following statements agree with the claims of the writer?

Write:

• TRUE if the statement agrees with the information
• FALSE if the statement contradicts the information
• NOT GIVEN if there is no information on this

9. The 2026 joint appeal was issued by the UN Secretary-General and the President of the ICRC.
10. The proposed regulatory regime would ban all autonomous weapons without exception.
11. A machine, in the writer's view, can reliably distinguish a surrendering combatant from a fighting one.
12. Autonomous weapons require enriched uranium for their operation.
13. No country has yet publicly fielded a fully autonomous weapon that selects and engages human targets.

Questions 14-15

Complete the summary below using NO MORE THAN TWO WORDS from the passage.

Campaigners call for an international treaty prohibiting autonomous weapons that target humans directly or that operate without (14) __________ human control, by analogy with treaties that banned chemical weapons and anti-personnel (15) __________.

答案与解析

← 上一篇 | 返回雅思焦点 | 下一篇 →