英伟达黄仁勋盛赞 OpenClaw 助力机器人技术发展 但权限管控必须得有NVIDIA's Jensen Huang Praises OpenClaw for Advancing Robotics, but Stresses Permission Controls Are Essential


AI free token/AI免费积分

IT之家 10 月 15 日消息,据科技媒体 TechRadar 今天报道,英伟达首席执行官黄仁勋近期在伦敦的 Build a Claw 活动认为,OpenClaw 将推动将来科技发展,特别是机器人领域。

英伟达黄仁勋盛赞 OpenClaw 助力机器人技术发展 但权限管控必须得有

黄仁勋首先指出,OpenClaw 是迄今为止最关键的模型之一,有助于多个行业进步。他对此解释称:“笔者已经看到人工智能迎来转折点,它开始变得实用。人们正在用 AI 做各种事情,从简单的软件编程,到担任营销代理、研究助理和实验室助理、药物研发,如今到处都有智能体。”

IT之家从原报道获悉,黄仁勋看到合作伙伴 Humanoid 的机器人后认为,OpenClaw 是“首个真正可信的智能体运行框架”。他认为 OpenClaw 就像外骨骼的大语言模型,应用在机器人领域可能带来许多好处。

他说道:“OpenClaw 刚推出时,我就觉得它是格外出色的数字机器人平台。它有点像具身智能,本质是为 LLM 给出了实时运行框架。电机、关节、车轮和机械臂都只是工具,最终要由电脑运行的智能体操控。”

“这种情况与互联网早期很相似。当时浏览器并未与计算机隔离,使用者访问网站时,这些 Web 页面可能会下载应用程序,并取得对底层硬件的完全控制权,而且还会向使用者的电脑推送它所说的「特殊软件包」。”

“最显而易见的应对方法就是建立容器,也就是隔离层,本质上是在操作平台内部再运行一个 OS。因而笔者开发了 OpenShell 平台。当你开发一个自主平台时,首先应该假设任何对象都不可信,仅授予最低权限。例如你让一个自主平台做煎蛋,就没必要给它电锯,它只需要一口煎锅。”

关于 AI 智能体失控的论调,黄仁勋比较乐观,他认为权限管控是必要的:“笔者会持续监控这些自主平台。如果机器人违反规则,或者做出超出权限范围的行动,平台就会发出警报。笔者可以决定是否让它闭嘴。”

本文来源:https://www.ithome.com/1/013/610.htm
若有版权问题,请来信告知:13951431913#139.com 或在右方提交评论。

IT House October 15th news, according to technology media TechRadar reported today, Nvidia CEO Huang Jenxun's recent Build a Claw event in London believes that OpenClaw will promote future technology development, especially in the field of robotics.

Huang Jenxun first pointed out that OpenClaw is one of the most critical models to date, contributing to the advancement of multiple industries. He explains: “I've seen AI come to a turning point and it's starting to become practical. People are doing everything with AI, from simple software programming to working as marketing agents, research assistants and lab assistants, drug discovery, and now there are agents everywhere.”

IT Home learned from the original report that Huang Jenxun saw the robot of partner Humanoid and believed that OpenClaw was "the first truly credible framework for agent operation". He believes that OpenClaw is like a large language model of exoskeletons, and its application in the field of robotics may bring many benefits.

“When OpenClaw first launched, I thought it was a fantastic digital robotics platform,” he says. It's a bit like being smart, essentially giving the LLM a framework to run in real time. Motors, joints, wheels, and robotic arms are just tools, ultimately controlled by a computer-driven intelligence. ”

NVIDIA's Jensen Huang Praises OpenClaw for Advancing Robotics, but Stresses Permission Controls Are Essential

“It's very similar to the early days of the internet. At that time, the browser was not isolated from the computer, and when the user visited the website, these web pages may download the application, take full control of the underlying hardware, and push what it calls" special software packages "to the user's computer."

“The most obvious response is to build containers, or isolation layers, that essentially run another OS inside the operating platform. Therefore, I developed the OpenShell platform. When you develop an autonomous platform, you should first assume that no object is trusted and only the lowest permissions are granted. For example, if you let an independent platform make fried eggs, there is no need to give it a chainsaw, it only needs a frying pan."

Regarding the argument that the AI agent is out of control, Huang Jenxun is more optimistic. He believes that permission control is necessary: "I will continue to monitor these autonomous platforms. The platform alerts if the bot breaks the rules or takes actions beyond its authority. I can decide whether or not to shut it up."

This article is from: https://www.ithome.com/1/013/610.htm For copyright issues, please write to: 13951431913#139.com or submit a review on the right.

暂无评论,快来发表第一条评论吧!No comments yet. Be the first!

📮 需求咨询📮 Inquiry