People in the artificial intelligence industry warned on Monday at a New York City Council hearing.
人工智能业内人士周一在纽约市议会听证会上发出警告。
They said this new technology could be dangerous, and they called for stronger rules to control it.
他们说,这项新技术可能带来危险,并呼吁加强监管。
Workers who recently left Anthropic, OpenAI and Google DeepMind said these companies have a reckless culture.
最近离开Anthropic、OpenAI和谷歌DeepMind的员工说,这些公司有一种鲁莽的文化。
They are racing to develop technology that they do not know how to control.
它们正在竞相开发自己不知道如何控制的技术。
Jacob Coxson is a former engineer at Anthropic. He said these companies have a startup mindset of "move fast and break things."
Anthropic前工程师雅各布·考克森说,这些公司抱着“快速行动、打破常规”的创业心态。
He believes this works for photo-sharing apps. But he says it is not right for building the most powerful technology in history.
他认为,这种做法对照片分享应用有效,却不适合构建史上最强大的技术。
He also said that on the current path, humans are more likely to lose control of AI.
他还说,按照目前的路径,人类更可能失去对人工智能的控制。
In the end, it could even lead to human extinction.
最终甚至可能导致人类灭绝。
The Trump administration is not worried about AI fears, and Congress has not taken action.
特朗普政府对人工智能恐惧不以为然,国会又没有行动。
So state and city governments are now thinking about how to regulate AI.
因此,州和市政府正在考虑如何监管人工智能。
Last month, Trump signed an agreement that lists voluntary steps companies can take.
特朗普上月末签署了一项协议,列出公司可采取的自愿措施。
This basically lets them regulate themselves.
这基本允许它们自我监管。
At the hearing, workers from OpenAI, Anthropic, Google and Meta also spoke.
在听证会上,代表OpenAI、Anthropic、谷歌和Meta的员工也作证。
They said their systems are already improving daily life. They also stressed that they care a lot about safety.
他们说,他们的系统已在改善日常生活,并强调他们高度重视安全。
Anthropic and OpenAI said their advanced AI products need rules.
Anthropic和OpenAI表示,其先进人工智能产品需要监管。
The company representatives were also open to some ideas from the city council.
公司代表也对市议会的一些建议持开放态度。
City Council President Julie Menin asked company representatives to measure AI's risk in the worst disaster situations.
市议会议长朱莉·梅宁要求各公司代表量化人工智能在最坏灾难性场景下的风险。
Morgan Dwyer from OpenAI said he was not sure about the chance of a disastrous mistake. He said it could be 1%, 10%, or 20%.
OpenAI的摩根·德怀尔说,他不确定发生灾难性错误的概率是1%、10%还是20%。
But he said that all of these levels are unacceptable.
但他说,这些水平都不可接受。
Menin criticized this answer as "careless at best."
梅宁批评这一回答“充其量是轻率的”。
The city council is considering several AI laws.
市议会正在考虑多项人工智能法规。
These laws include asking companies to share information about their AI tools.
这些法规包括要求公司披露人工智能工具信息。
They also include a rule that the city's AI systems must get third-party approval.
它们还包括要求该市的人工智能系统获得第三方批准。
In addition, they let whistleblowers share in fines paid by companies that break the law.
此外,它们允许举报人分享对违法公司的罚款。
Local officials like New York Governor Kathy Hochul have also taken action. They are pushing for AI reporting and transparency standards.
纽约州州长凯西·霍楚尔等地方官员也已采取行动,推动人工智能的报告和透明度标准。