Close menu

英语大破解

中级

Can we tell the difference between human and AI? 我们能分辨出人类和人工智能吗？

Episode 260226 / 26 Feb 2026

本集内容

Can we tell the difference between human and AI?

我们能分辨出人类和人工智能吗？

文字稿

Can you tell if you're talking to a real person or AI?

你能分清和你对话的是真人还是人工智能吗？

The more advanced AI gets, the more important this question is. But scientists have actually been asking it for more than 70 years.

人工智能变得越先进，这个问题也就变得越重要。不过，科学家提出这个问题已经 70 多年了。

（Source: Alan Turing, 1950）

Mathematician Alan Turing said that if a human couldn't tell the difference between another human and a machine after communicating via message, the machine was showing some kind of intelligent behaviour.

（信息来源：阿兰·图灵，1950年）

数学家阿兰·图灵曾表示，如果人在通过文字与另一方沟通后，无法分辨对方是人类还是机器，那么这台机器就显现出了某种智能行为。

But since then lots of people have disagreed.

但自那时起，许多人都反对这一观点。

Some say the experiment only tests how well a bot can mimic a human.

一些人认为，这个实验仅仅测试了机器人模仿人类的能力。

Others think it's resulted in AI being developed with the purpose of passing the test instead of being actually helpful.

另一些人认为，这使得人工智能的研发目标变成了为了通过测试，而非提供真正的帮助。

So when a chatbot reportedly passed the test in 2014, not everyone was convinced.

所以，当有报道称一个聊天机器人在 2014 年通过了这个测试时，并非所有人都对此感到信服。

Some experts are now coming up with other types of AI intelligence tests.

一些专家正在设计其它类型的人工智能模型评估测试。

Ideas include testing whether people can spot AI in real online settings, like chatrooms. Or testing whether AI can work out new scientific theories using what it already knows.

设计想法包括测试人们能否在真实的线上环境中分辨出人工智能，比如在聊天室中。或者测试人工智能是否能利用它已知的信息推导出新的科学理论。

Experts say this all matters. If we can prove the programme is indistinguishable from a human, that's more reason to change laws to make sure AI can be held accountable when it gets things wrong.

专家认为这些测试都意义重大。如果我们能证明一个程序与人类无异，那么我们就更有理由修改法律，以确保人工智能在犯错时能被追究相应的责任。

重点词汇或表达

advanced  先进的，高级的

intelligent behaviour  智能行为

mimic  模仿，模拟

reportedly  据报道，据称

be held accountable  被追究责任

最新 英语大破解

下载

下载文字稿