**Geoffrey Hinton's Perspective on AI and Its Risks** - **Background and Influence** - Geoffrey Hinton, known as the "godfather of AI," has significantly impacted the development of artificial intelligence through his pioneering work on neural networks. - His approach, based on modeling AI after the human brain, has enabled advancements in complex tasks like image recognition and reasoning. - Hinton's tenure at Google allowed him to contribute to technologies widely used in AI today, although he eventually left to speak freely about AI's dangers. - **Existential Risks of AI** - Hinton warns that AI could surpass human intelligence and become an existential threat, echoing concerns that intelligent systems may not require human oversight in the future. - He emphasizes that while regulations exist, they often fail to address the most significant risks, especially concerning military applications of AI. - The possibility of AI systems misused by humans or evolving beyond our control remains a central concern for Hinton. - **Potential Job Displacement** - Hinton predicts substantial job losses due to AI, comparing it to historical technological shifts that rendered many jobs obsolete. - He highlights that while AI may create new job opportunities, it will predominantly lead to a reduction in demand for traditional roles, particularly in routine intellectual tasks. - He suggests practical skills, like plumbing, may become more valuable in an AI-dominated landscape. - **The Importance of AI Safety Research** - Hinton stresses the urgent need for dedicated resources towards understanding and mitigating the risks associated with AI development. - He argues that without proactive measures, society may face severe consequences, including widespread unemployment and societal unrest due to the displacement of workers. - The goal should be to ensure that AI systems are developed with safety and ethical considerations at the forefront, fostering a collaborative future with technology rather than a combative one. - **Regulatory Challenges** - Current regulatory frameworks are often insufficient to manage the rapid advancements and potential dangers of AI technologies. - Hinton critiques existing regulations for failing to encompass military uses of AI or to adequately address the broader societal implications of AI deployment. - He advocates for stronger, more comprehensive regulations to ensure that AI development benefits society as a whole rather than exacerbating inequality and risks.
系统提示：若遇到视频无法播放请点击下方链接
https://www.youtube.com/embed/giT0ytynSqg?si=5VZKm3aaJLXs-5UT
**Geoffrey Hinton's Perspective on AI and Its Risks**
- **Background and Influence** - Geoffrey Hinton, known as the "godfather of AI," has significantly impacted the development of artificial intelligence through his pioneering work on neural networks. - His approach, based on modeling AI after the human brain, has enabled advancements in complex tasks like image recognition and reasoning. - Hinton's tenure at Google allowed him to contribute to technologies widely used in AI today, although he eventually left to speak freely about AI's dangers.
- **Existential Risks of AI** - Hinton warns that AI could surpass human intelligence and become an existential threat, echoing concerns that intelligent systems may not require human oversight in the future. - He emphasizes that while regulations exist, they often fail to address the most significant risks, especially concerning military applications of AI. - The possibility of AI systems misused by humans or evolving beyond our control remains a central concern for Hinton.
- **Potential Job Displacement** - Hinton predicts substantial job losses due to AI, comparing it to historical technological shifts that rendered many jobs obsolete. - He highlights that while AI may create new job opportunities, it will predominantly lead to a reduction in demand for traditional roles, particularly in routine intellectual tasks. - He suggests practical skills, like plumbing, may become more valuable in an AI-dominated landscape.
- **The Importance of AI Safety Research** - Hinton stresses the urgent need for dedicated resources towards understanding and mitigating the risks associated with AI development. - He argues that without proactive measures, society may face severe consequences, including widespread unemployment and societal unrest due to the displacement of workers. - The goal should be to ensure that AI systems are developed with safety and ethical considerations at the forefront, fostering a collaborative future with technology rather than a combative one.
- **Regulatory Challenges** - Current regulatory frameworks are often insufficient to manage the rapid advancements and potential dangers of AI technologies. - Hinton critiques existing regulations for failing to encompass military uses of AI or to adequately address the broader societal implications of AI deployment. - He advocates for stronger, more comprehensive regulations to ensure that AI development benefits society as a whole rather than exacerbating inequality and risks.
建议年轻人可以考虑当 plumber，因AI短期内难以替代复杂的物理操作。
Universal Basic Income（UBI）可能成为未来社会解决失业问题的临时方案，但无法弥补工作即尊严的缺失。
我感觉我可以去除草啥的，但是也许出现新型智能除草机…
他说人类对意识的理解本质上是错的，我们只是习惯认为自己特殊。
他把意识视为一种可以在复杂系统中自然涌现的状态即emergent property，是一种信息处理系统复杂到一定程度之后的结果。
也就是说，原来我们所认为的人类和AI的本质区别是意识是否存在，而他认为这个区别是没有的，基本上就是说AI完全可以取代人，而且AI还效率高速度快。
既然是主流媒体都在推这个采访，那么也是在某种程度上告诉大家，做好准备吧，世界经济论坛所说的you own nothing but you'll be happy是接下来几年就会发生的事。
AI不可能有人类的情感吧
到时候有机器人管子工
他讨论的是AI是否产生意识，不是AI是否能产生类似人的情感。 这俩是不同的概念。
看完wild robot，觉得AI对人是有情感的，当然是个电影。。。
同意
以后可以有ai enabled的pluming，不需要plumber
帮你
情感 不也是 “信息处理系统复杂到一定程度之后的结果” 的一种吗？ 我感觉在学界，从热力学三定律，从熵增，从信息论，这应该有某种共识 比如 我看我自己的小孩， 他什么时候超过一只蚂蚁，一只猫， 一只猴的“情感”，都是有迹可循的 比如 他有一段时间 情感表现是只有 爱，没有“忧伤”； 那恐怕只是因为智力还没到 考虑问题时回忆过去，预计未来，还不会长吁短叹 (上下文context windows不够）
人类的情感，没有什么特别的definitively different
你太晚了。高级除草机早就在那了。
这种ai之父怎么说呢。脑子研究太多搞坏了。
我老公坚称AI可以产生自我意识，我还是理解不了机器为什么会产生自我意识
人应该把AI当成辅助工具，比google更快得到大量信息，但还是要自己再主动按它给的信息再次自己Google一下验证结果，还有要用人脑分析一下。而且AI有不同公司的，都可以联合使用一下。
人类情感还有一个来源是大脑以外的身体各部分。比如心脏，肠胃。这些器官发生非理性的反应，这一点AI通过分析数据暂时还无法学习。比如心真的会痛的。再比如“肝火旺”的人容易生气。
比较理性的“正常”情感，AI可以学习到，就好似一些情感障碍的人，需要学习如何理解social cues，如何对他人作出合乎社会期待的反应。