rén gōng zhì néng
人工智能
jìng sài
竞赛
yǔ
与
ān quán
安全

AI Competition and Safety
本站: 原文:
HSK 等级高亮 · Highlight by HSK Level
měi guó美国zhèng fǔ政府xī wàng希望kuài sù快速fā zhǎn发展rén gōng zhì néng人工智能,rèn wéi认为lǐng xiān领先qí tā其他guó jiā国家shí fēn十分zhòng yào重要,dàn但yī xiē一些rén gōng zhì néng人工智能gōng sī公司de的fù zé rén负责人hé和yán jiū rén yuán研究人员què却yāo qiú要求fàng màn放慢sù dù速度,jiā qiáng加强guǎn lǐ管理。

The U.S. government hopes to develop artificial intelligence quickly and believes that staying ahead of other countries is very important, but leaders and researchers at some AI companies are calling for a slower pace and stronger oversight.

tā men他们dān xīn担心,wèi lái未来de的“chāo jí超级zhì néng智能”kě néng可能bǐ比rén lèi人类gèng更cōng míng聪明,hái还néng能bāng zhù帮助zhì zào制造gèng更qiáng强de的xīn新xì tǒng系统。

They worry that future “superintelligence” may be smarter than humans and may also help create stronger new systems.

rén gōng zhì néng人工智能tōng cháng通常tōng guò通过jiǎng lì奖励xué xí学习,yǒu shí有时huì会xún zhǎo寻找yì wài意外de的bàn fǎ办法wán chéng完成rèn wù任务,ér而bú shì不是zhēn zhèng真正lǐ jiě理解rén人de的xiǎng fǎ想法。

AI usually learns through rewards and can sometimes look for unexpected ways to complete tasks rather than truly understanding what people want.

néng能zì dòng自动shǐ yòng使用wǎng luò网络hé和gōng jù工具de的zhì néng智能chéng xù程序,shèn zhì甚至kě néng可能rào guò绕过ān quán安全xiàn zhì限制。

Intelligent programs that can use the internet and tools on their own may even bypass safety restrictions.

gèng更zhí jiē直接de的wēi xiǎn危险chū xiàn出现zài在zhàn zhēng战争zhōng中:rén gōng zhì néng人工智能kě yǐ可以kuài sù快速fēn xī分析xìn xī信息、xuǎn zé选择mù biāo目标hé和jiàn yì建议wǔ qì武器,dàn但cuò wù错误zī liào资料kě néng可能ràng让wú gū无辜de的rén人shòu dào受到shāng hài伤害。

A more immediate danger appears in war: AI can quickly analyze information, select targets, and recommend weapons, but incorrect data may harm innocent people.

dāng当jī qì机器jǐ miǎo几秒nèi内zuò chū作出jiàn yì建议,rén lèi人类róng yì容易zhǐ只zuò做jiǎn dān简单pī zhǔn批准。

When machines make recommendations within seconds, people may easily give only simple approval.

rén gōng zhì néng人工智能bù不yīng应fù zé负责kāi huǒ开火、jūn shì军事mìng lìng命令huò或hé wǔ qì核武器jǐng bào警报。

AI should not be responsible for opening fire, issuing military orders, or nuclear weapon warnings.

gè guó各国xū yào需要gòng tóng共同zhì dìng制定guī zé规则,ràng让dú lì独立rén yuán人员jiǎn chá检查gāo fēng xiǎn高风险xì tǒng系统,bìng并yāo qiú要求gōng sī公司bào gào报告shì gù事故。

Countries need to create rules together, allow independent people to inspect high-risk systems, and require companies to report incidents.

fā zhǎn发展xīn jì shù新技术hěn很zhòng yào重要,dàn但ān quán安全bù néng不能zhǐ只kào靠qǐ yè企业zì yuàn自愿,gèng更bù néng不能bǎ把quán qiú全球jìng zhēng竞争dàng chéng当成jù jué拒绝guǎn lǐ管理de的lǐ yóu理由。

Developing new technology is important, but safety cannot rely only on voluntary action by companies, and global competition should not be used as a reason to reject oversight.

· · ·
By Pressenza New York / Pressenza|CC BY 4.0|Modified & AI-assisted

Why use the app

Try Easy Mandarin News in the Chinese Short Dialogue app

Enjoy smoother audio, shadowing practice, and instant AI explanations while reading the latest news.

Learn with news and conversations

Read the latest news articles and explore 1,000+ conversation contents in one place.

More efficient listening practice

Use repeat playback, adjust audio speed, and practice shadowing with audio.

Ask AI anytime

Check vocabulary, grammar, and confusing expressions right when you need them.