雅思阅读 200: A Person or a Product? — The uneasy legal status of intelligent machines（人还是产品？智能机器尴尬的法律地位）

改编自 LawTalk / Frontiers in Robotics and AI / EU AI Act（2021-2026年）。雅思阅读 Section 3 难度，约 1050 词。 素材来源：https://lawnews.nz/technology/beings-like-us-consciousness-personhood-and-the-legal-status-of-artificial-intelligence/

Reading Passage

A. The word "person", in law, has never simply meant a human being. Corporations have been treated as persons for centuries — they can own property, sign contracts and be sued, even though they have no body, no breath and no life. Ships, churches and, in some ancient systems, even gods have been granted a legal standing that lets the law act on them, fine them or bring suit against them. Personhood, in other words, has always been a practical device rather than a biological description: it is whatever the law chooses to recognise as a single point of responsibility. So when lawyers ask whether an intelligent machine could one day be a person, they are not necessarily asking whether it can feel. They are asking a narrower, practical question: can we hold it, or the law around it, in such a way that responsibility for its actions becomes clear? This is the same question courts have always faced when a new kind of actor appears on the scene. As artificial intelligence systems begin to act with less and less human oversight, that question has moved from science fiction onto the desks of parliaments. The more such systems decide for themselves who gets a loan, who sees an advert or who is hired for a job, the more urgent it becomes to name the responsible party.

B. The first major foray into the debate came from Europe. In 2017, the European Parliament adopted a resolution, proposed by the Luxembourgish Mady Delvaux, that floated the possibility of a specific legal status for the most sophisticated autonomous robots. The idea, in its strongest form, was that such machines might become "electronic persons" — entities that could be assigned rights and duties, and could in principle be made to compensate for harm they caused, perhaps by holding a fund of their own. The reasoning was pragmatic rather than sentimental. If a machine acts on its own and injures someone, a gap opens between the human programmer, the owner and the victim; no single person may be at fault, and yet someone must answer to the person harmed. Making the machine itself a bearer of liability looked, to some, like a tidy way to fill it. The proposal was exploratory, but it sent a jolt through both technical and legal circles. Engineers who built such systems worried that a machine could not really be blamed, while lawyers wondered what it would even mean to fine a piece of software.

C. The jolt was answered almost immediately. More than two hundred experts from fourteen countries signed an open letter warning against the very notion. Their core argument turned on what moral responsibility actually requires. To deserve praise or blame, they contended, an agent must possess subjective experience — intentions, desires, an inner life of some kind. Present artificial intelligence, however sophisticated its outputs, does not experience anything; it processes without feeling, however much its prose may suggest otherwise. When a chatbot writes a moving sentence, no one is home inside the machine to feel it. To grant such a system the status of a person, the critics argued, would not elevate the machine but dilute the law: it would let the humans behind it hide behind a fictitious entity, obscuring who actually chose what, who profited and who should pay. Personhood, on this view, is not a label you confer on a clever tool.

D. Europe, in the end, chose the critics' path. When the European Union adopted its landmark AI Act in 2024 — the first comprehensive law of its kind anywhere — it explicitly rejected the idea of "electronic personhood". Advanced systems are regulated not as persons but as products, graded by the risk they pose: a narrow set of uses is banned outright, high-risk systems face strict controls, and the rest are left comparatively free. The choice was deliberate, and it marked a retreat from the 2017 flirtation. Today, no AI system anywhere on Earth holds legal rights; courts on every continent continue to treat even the most capable software as property owned by someone. The autonomy that looks so striking on a screen is, legally, still just a sophisticated thing a person made, and the law can always reach past it to the human who built, sold or deployed it. In the meantime, the burden of risk is expected to fall on those humans rather than on the machine.

E. And yet the underlying problem that prompted the 2017 proposal has not gone away; it has only been postponed. The product model works cleanly when a machine fails in a predictable way, but modern systems can act in ways their designers did not foresee and cannot fully explain. When such a system causes harm, tracing responsibility back through layers of training data, model and deployment becomes genuinely hard, and a plaintiff may face a wall of plausible deniability. Some scholars argue that a narrower, functional status — borrowed from the old idea of corporate personhood, and limited to bearing liability rather than enjoying rights — may yet prove necessary. Whether that day ever arrives depends less on a philosophical verdict about consciousness than on a practical one about accountability. It is possible to imagine a future in which a machine is held liable the way a company is — a legal fiction with a bank account behind it — without ever granting it a mind or a conscience. The law, historically, has granted personhood not to whatever seems most human but to whatever the world needs to hold responsible. Machines may become persons for the same unromantic reason: because someone has to pay. It is a prospect no philosopher demanded but that the practical law may yet choose.

Questions 1-4

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

List of Headings i. The European proposal that startled the field ii. Why "person" need not mean a human being iii. The expert backlash and the case for consciousness iv. Europe's final choice: a product, not a person v. The unresolved liability gap vi. The history of corporate law vii. How AI systems are trained

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. Why does the writer say "person" need not mean a human being? A. Because only humans can own property. B. Because entities such as corporations have long been treated as persons in law. C. Because robots are already human. D. Because the law has never recognised animals.

6. What was the practical aim of the 2017 "electronic personhood" proposal? A. To give robots the vote. B. To make sophisticated autonomous machines bear liability for harm they cause. C. To ban all artificial intelligence. D. To declare that machines can feel pain.

7. Why did more than two hundred experts oppose the proposal? A. They believed AI systems already had full consciousness. B. They argued moral responsibility requires subjective experience that current AI lacks. C. They wanted robots to be taxed more heavily. D. They feared the proposal was too slow.

8. How does the EU AI Act of 2024 treat advanced AI systems? A. As legal persons with full rights. B. As products graded by risk, and explicitly not as persons. C. As citizens of the European Union. D. As unregulated technology.

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. Corporations have been treated as legal persons for centuries.
10. The 2017 resolution was proposed by Mady Delvaux.
11. Every AI system in the world currently holds legal rights.
12. The EU AI Act of 2024 embraced "electronic personhood".
13. Mady Delvaux held a doctorate in computer science.

Questions 14-15

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

The 2017 proposal suggested that the most autonomous machines might become "electronic (14) __________", but the 2024 EU AI Act instead regulates AI as a (15) __________ graded by its risk.

答案与解析

← 上一篇 | 返回雅思焦点