Patrick John

[Patrick John @Patrick_theon]

@claudeai block personal user is a stupid decision.

[Patrick John @Patrick_theon]

@ClaudeDevs There are many personal users can contribute to this, but are all blocked. It's really a stupid decision that limit yourself.

[Patrick John @Patrick_theon]

@claudeai As leading AI lab in the world, it should keep the equality for every one access to the ability, or at least paid users. Malicious distillation should be smartly detected instead of cutting off all users.

[Patrick John @Patrick_theon]

This is what we normally called as society efficiency. This is the most important that we need to consider when we evaluate a country's or an organization's development potentials.

[Patrick John @Patrick_theon]

The core content of human society are human themselves. The essence drives society progressing is human creativity, knowledge, but all these shall be effective through human's intelligence and time. Although human's intelligence is unlimited but their time is limited.

[Patrick John @Patrick_theon]

For a person himself, every choice he made is based on his own conditions, reconginitions and experiences. But when it comes to a society level, this normally is not decided by one person, while most of the deicisons are made or orchestrated by the governance and institute.

[Patrick John @Patrick_theon]

So we shall be very picky on our choices, like careers, study, friends, couples, etc.

[Patrick John @Patrick_theon]

@AnthropicAI When you can stop blocking user’s account? The blocking can only just hurt the normal users instead of the vicious.

[Patrick John @Patrick_theon]

@Tsla99T 也许你真的技术很牛逼，但是政治上不够成熟（local人脉、资源、资金、思想觉悟），也不行。

[Patrick John @Patrick_theon]

@Tsla99T 对于很多在美国初有成就的年轻人想回国闯荡，还是要谨慎，你即将面对的不是单纯的技术或产品难题。你即将面对的是很不同的平台和环境。稍不留意公司、文化、团队就会教你做人。比如敬酒该怎么敬、汇报怎么摸透领导的心思、第一时间是要搞出噱头，对得起自己人才的title，而不是什么底层技术攻关。

[Patrick John @Patrick_theon]

@linyiLYi 虽���我从来不是技术悲观主义者，但目前看到的这个竞争态势，并不是一个开放包容的竞争。虽然大家各���都有安全的理由筑起高墙，但是这个安全理由是否足够透明公开，是否会被滥用，普通民众并不知道。从这个角度来说，也许主权AI是必要的，也是必需的。不为别的，就为达到类似核威胁的恐怖平衡一样。

[Patrick John @Patrick_theon]

@linyiLYi 打磨性能的能力会随着训练数据和算力的提升而不断提升，而其中最重要的训练数据尤其是涉及到安全、商业机密等又是未来可以区别大模型能力的关键。如果哪个组织愿意把自己的私有数据用来大模型的私密训练，我相信会让大模型的能力在某个领域更上一层楼，但是由于现在大模型公司与私有数据公司往往不同源

[Patrick John @Patrick_theon]

@linyiLYi 不过在这个各方利益争夺和不断筑高墙的过程中，往往被甩下战车的或被随意碾压的就是普通民众。他们没有资本，没有能力，只能任由大模型和有能力建设大模型���力的组织摆布。按照这个逻辑往下发展，也许未来AI的���代，会是一个大资本或权力驱动的AI主导的反乌托邦式的世界。

[Patrick John @Patrick_theon]

@linyiLYi 随着大模型公司彼此之间为了保护自己的利益而不断筑起高墙（不管以什么名义），而大家在AI的竞赛中又都不愿意落后，自然主权AI的概念也许就是自然竞争演化的产物。

[Patrick John @Patrick_theon]

@linyiLYi 导致这个数据壁垒一直难以打破。也许未来等到算力足够便宜以后，每个独立的组织都可以在公共模型的基础上进行基于私有数据的私有模型训练，从而不断加深自己在既有行业的壁垒深度。但这也许就自然演化出来了黄仁勋提出的“主权AI”的问题。

[Patrick John @Patrick_theon]

@AnthropicAI That sounds great! But it is not so exciting for users outside of America, since there are lots of ristrictions.