非花 retweeted
这攻击太密集了,一天发十八篇大字报。。。文革批斗,还给子女机会跟父母划清界限。。。这就是灭门吧。。。连坐。
黄仁勋女儿留任引发担忧:脾气暴躁,曾导致多名高管离职。黄敏珊此前是父亲事实上的幕僚长,她脾气火爆、作风强势,英伟达员工称有十多名关键高管因为不喜欢与她共事而离开公司。她继续留任给临时CEO带来了直接挑战。
cn.nytimes.com/business/2026…
以前的美国:舆论抹黑完派出军队,赢得一场战争
现在的美国:舆论抹黑完手足无措,输到莫名其妙
China taps US AI models to train homegrown rivals as chip curbs bite reuters.com/world/china/chin…
非花 retweeted
Replying to @xeno_swift @Logs_importer
民间虽少有人提,
但官方、学术乃至教育,孙文先生的国父地位从始至终都是被肯定、传承和发展的。
民族、民权、民生
军政、训政、宪政
以党治国、民族统一为先
平均地权、节制资本、《建国方略》
“10万英里铁路、三大港口”、三峡大坝、平陆运河等等
中华人民共和国从始至终,都是孙文思想传承、发展的践行者。
飞机停飞了,银行宕机了,但IRS总能工作正常😥😥😥
🇺🇸 Chase Bank went down for thousands of customers with most of them unable to log in online.
Users are now reporting problems at Bank of America, Wells Fargo and U.S. Bank too and there's no word yet on what's behind it.
First the flights got grounded and now the banks are down. What's going on?
Writer: Daniyal
2
13
863
非花 retweeted
我发现我疯狂爱上了字节跳动的这个女生😳曾妍
1997年生,2021年9月校招进AI Lab做算法工程师,2026年3月被曝从3-2升到4-2,一年连升两级。外界估算,这个级别算上工资、年终奖和股权,一年总包可能到500万以上。
如果能娶到她,是不是算男生也拿到大结果了?
原本预测IRS系统升级,完全没想到,竟然是美国主要银行正面临了广泛服务中断!部分Chase、美国银行和富国银行客户无法访问其账户。Chase承认目前存在问题,因为数千起中断报告在网上涌现。
Reliable sources confirm the IRS is completing its full upgrade to an AI-controlled platform tomorrow, November 15th. This comes after Microsoft’s wildly successful upgrade that briefly turned the country’s email and Teams into a mess, and the FAA’s cutting-edge AI system that innovatively grounded all of America. The IRS’s action plan is both brave and responsible. (SATIRE_AI)
Replying to @Faewha
West has become more corrupt than ever. AI is the biggest threat to "Western Democracy" in their eyes.
I called this a while ago but China has already won the 'AI race'. Western frontier models do not allow consumers to use the full capability of the models they serve. They are already trying to censor what people can and cannot build. If you are working on a project that is deemed competitive by themselves or their partners the model will block or simply sandbag the development of that work. This has been documented since Opus 4.6.
People think the benchmark scores they see are everything – but the frontier AI companies are accelerating horizontal capabilities (game design etc) and throttling all vertical capabilities in the name of safety. People think they are getting much better models but it's an illusion. Try working on a project that competes with them and see how far the 'frontier' takes you. Safety doesn't concern them or their partners though. They want to control what AI everyone else has access to and it has to be under their terms.
This is the reality. If that's the product, and it costs 10-100x more than a self-hosted open weight Chinese model then they simply cannot win the AI race. They can win the AI race behind closed doors only – and they will – but watch them only serve their elite partners and government. US AI is a snake eating its own tail.





























