dvduval

@jun_song It’s nice when I wanna go to some other activities besides sitting in front of the computer and it will just chime in when something has been completed and then I can follow up and tell it what I want to do next or just talk to her about what’s going on.

Can someone explain to me why everyone is calling Claude/Codex voice revolutionary, and how it’s actually any different from speech-to-text that’s been around for over 10 years?

@argofowl I’ve been yelling at it all night. Unfortunately, it doesn’t say anything back.

is there a single person who got access to the new voice mode in codex?

@jun_song The people with the big money will control the energy and the chips and the data centers to decide how that intelligence is used. Open source does not solve this problem.

I’ve been a hardcore advocate for AI acceleration for a long time. Looking at my posts, a lot of people might assume I’m a decel. I read Ray Kurzweil’s "The Singularity Is Near" almost 15 years ago back when I was a student, and I’ve spent years studying and believing that this is the future. I kept talking about the Singularity to people around me, but until just 2 years ago, they treated me like a conspiracy theorist. Now, we are literally standing at the threshold of that Singularity. However, I’ve always believed that the hype overblown by frontier labs to secure funding, combined with their attempts to shut out open source and build monopolies, is fundamentally wrong—so I called it out. Yeah, when you point out what frontier labs are doing wrong, a lot of people walk away. They just want hype and acceleration, and they don’t want to face the truth. Sometimes when I read posts by others and see replies from people who blocked my account, it honestly hurts. But I’m going to keep pushing for what I believe is right. I truly believe Open Source AI is the only thing that will drive real acceleration and stop monopolies. Thank you for paying attention to this issue.

@thomasunise @mkratsios47 Any company has the right to make their content private. If something is published publicly on the Internet, there must be the expectation that other people will use it. If there is a direct copyright infringement, that is something that can be addressed.

@mkratsios47 Anthropic stole from every independent publisher on the Internet

We have information that Moonshot AI distilled Anthropic’s Fable for the development of its K3 model. To do this they developed a sophisticated internal platform to conduct large scale distillation against U.S. models, allowing them to quickly switch between multiple methods of access to avoid detection. Moonshot AI has also acquired GB300-equipped servers and has accessed GB300s in Thailand, likely to train its AI models.   The United States strongly supports the free and fair development of AI, including a thriving competitive ecosystem that spans frontier models, specialized systems, open-source frameworks, and open-weight models. Legitimate AI distillation used to create smaller, more efficient models plays a vital role in this open innovation ecosystem. However, large-scale, covert industrial distillation aimed at stealing proprietary U.S. technology and undermining American research is unacceptable.

@bindureddy I think which key here is the scale level were involved over 20,000 accounts and millions of queries. I don’t blame them for wanting to protect themselves.

Sigh, US companies and govt complaining that China distilled model outputs is SUPER LAME Labs routinely distill model outputs from each - Meta, Grok, and others do it all the time TBH, even Anthropic trained their models on large agentic loops run by their users - they took someone else's data So what exactly is the big deal?

@jun_song Good, I hope you stop talking about them too.

Anthropic literally stole data from all of humanity. I’m never spending a single penny on Claude ever again.

@Prathkum Ha ha that’s hilarious, as China is going to work building closed models

The best thing about the Chinese AI labs and their open weight models is that they don’t create unnecessary hype around every release like Anthropic does. They just ship capable models that cover almost everything most of us actually need. Fable 5, on the other hand, feels like a token eater that is only a slight improvement over Opus 4.8.

Yes, I basically agree with that notion. There’s nothing going to happen in the short term to America. But things could happen slow slowly in the long-term. But if you look at a place like China, they’re losing population right now.

@quxiaoyin There’s a difference between banning models running on Chinese servers and banning open source.

Banning Chinese open-weights means Chinese labs will continue to distill from U.S models, but U.S labs cannot distill from Chinese models. - Data acquisition cost for Chinese labs will remain low, sustaining their already thin margins; - Data acquisition cost for U.S labs will reach all time high, forcing them to increase their pricing even more.

@HealthRanger Yes, China will have access to the business data of companies that choose to put on Chinese servers for those countries. And everyone will have access to anything that open source and can be run on US servers.

If Trump moves forward with his reported plan to ban open source AI models out of China, then: - Brazil will have access to frontier intelligence models, but the USA won't. - Russia will have access to frontier intelligence models, but the USA won't. - India will have access... - Turkey will have access... - Pakistan will have access... - Vietnam will have access... ... etc. But U.S. companies will be forced to use government-approved, dumbed-down, nerfed AI models that are intentionally retarded and incapable of doing much. So every U.S. corporations of any real size will FIRE their U.S. tech workers, and they'll set up new offices in Brazil, Argentina, Vietnam, India, Pakistan, etc., where they can actually use the world's best AI models (from China), and that's how you'll see a mass exodus of tech jobs away from the USA while the country also loses the AI race to superintelligence. I call this the "MADA" plan - Make America Dumb Again. I'm sure Anthropic and OpenAI will push for it.

@thsottiaux Hoping you’re just getting started. We gotta go past 10 for sure.

10M! New day, new usage reset for paid users of Codex and ChatGPT Work. Lands in the next hour. Enjoy.

@HealthRanger If we wanna talk about something dumbed down, try asking the Chinese models about Taiwan or Tibet or criticisms of the leader of China.

Serious AI question. If the White House is going to demand that Americans are only allowed to use dumbed-down U.S. AI models while banning access to China's superior open source AI models, who will pioneer the backpropagation of model retardation through all 48 layers?

@DavidLe76335983 I think the communist part is when you start asking any of the Chinese models about Taiwan or Tibet for example. Or ask them to tell you about the major criticisms of the leader of China. They won’t do it.

Must be panic time when you have to call winning AI model (open source for all to see) as communism

@IamChrisTurner @bindureddy It just depends how you define what America is. Do you mean the government? Or does that include private interests? If you’re including the influence of private America is by far the most powerful in the world.

I never said the US is weak, I said it doesn’t have as much control as people think. You claimed the US dominates because it controls everything. I’m pointing out that it doesn’t. The dollar’s dominance is artificial and it will lose its status the moment its role as the default currency for oil changes. Fortunately, the US still has the world’s strongest military, and it isn’t afraid to use it to maintain its position. However, it doesnt control mich outside markets. I love America but its not the centre of the earth like its people believe. You have beautiful national parks though, I’ll give you that.

Wow - woke up to some really bad news The nightmare scenario is coming true - the US may ban open source models Europe and China will dominate the global AI market and kill the US - If this happens

@ziwenxu_ The really powerful models will have a strong infrastructure supporting them for quite some time. That’s why China is also developing closed models.

@phpld That’s what I’m think about too. If it’s open source they can’t really block it at all.. Will be weird trying to block people to download it

JUST IN: Trump officials are reportedly weighing blocking U.S. access to top Chinese AI models. Sit with that for a second. Not "outcompete." not "build something better." block. You don't call a national security meeting over the models no one uses. You call one over the model that's beating you. A year ago Kimi, Qwen and DeepSeek were the cheap thing you reached for when the OpenAI bill got scary. Now one of them is why there's a meeting on the calendar. A ban isn't a strategy. it's a scoreboard BTW

@macastel3 No, open source models will always be available. But what the US will block is access to Chinese servers that are serving up the models from China. Even China is building closed source models now.

A great firewall seems inevitable for US (modeled in Chinese one) If the Chinese models flood the market also the capex and stock market will crash It’s a short term fix but maybe the only one left

@IamChrisTurner @bindureddy Go take a look at the largest market in the world, and which country’s money supply is used throughout the world and then come back and tell me the US is weak.

@phpld @bindureddy Actually, the US control very little. In fact, the only time we hear about the US is when the orange man is making unpredictable moves. In Europe, what they do has little influence. I love the America but please do sit back down.

@chamath No, it won’t be like that at all because American corporations will have their own models they host them themselves. They will pay the cost of hosting. Nobody wants to host their stuff in China.

The leading AI has already forked into two options. A: Closed source American that costs $26-56 per 1MM tokens. B: Open weight Chinese that costs $0.50-1 per 1MM tokens. If you force American companies to spend 50-100x more than their competitors abroad here is what will happen: 1. American companies spending 50-100x will at some point become financially impaired and the stock market will crater. This is the equivalent of the US Government saying we can only buy oil at $800/barrel even when the free market sells the same oil for $80/barrel. 2. In the short term, the revenues of the Closed American Labs may remain hi. But then their revenue will also crater because their customers in the US will be impaired and their customers abroad will have already flipped to cheaper solutions. This would be a terribly self-defeating form of intervention if it were to happen.

@APompliano So they can block access to Google on Facebook, but we can’t block their apps?

Banning Chinese models is the most un-American thing we could do. Stop complaining and start competing. People won't use Chinese models if American companies build something better.