861–870 of 965 posts
Datapoint from the cheap end of the market: I run local models on a couple of Orange Pi SBCs and a decade-old Optiplex with no GPU. What runs usably on that class of hardware is almost entirely Chinese open weights — Qwen's MoE builds (35B total, ~3B active) are the only thing that gives me acceptable speed on CPU, with Gemma as about the only western exception. I evaluated Kimi too and ruled it out purely on size. W…
Is Gemma worse?
Earlier quoted context omitted.
Not sure I understand that position. Unless we're talking about a scenario in which one is using an outdated model along with no grounding (which, imo, PEBKAC), why would the model be pinning insecure libraries? If it does have grounding, and can therefore see that it's introducing vulnerabilities to the code it's generating, yet does so anyway... I suppose we could invoke Hanlon's razor, but if the model is that inc…
The model is conditioned / pretrained to use a particular version of a library. The model doesn’t know why — it was just taught to use that version. To the model, it is just some insignificant detail in the grand scheme of things. It’s just like how a model would intuitively favour using English without explicit instructions — it’s not trying to sabotage other cultures, it’s just what it does.
Also, in that case, there would likely be activations indicating that it is favoring a specific version. If that's an insecure version, sure that'd be suspicious... but again, you're only going to be able to verify that's what's happening in an open model.
Maybe you can illustrate a realistic scenario in which that would be a problem, otherwise I don't really understand what your point is in this context.
Earlier quoted context omitted.
You two are talking about different things. You are pointing out that OpenAI can cancel user's accounts for almost any reason, and nobody can really force them to serve customers that they suspect are distilling their models. That's one thing. The GP is saying the government can make laws to make terms against distillation unenforceable. Without such laws, if you signed an agreement with OpenAI pinky swearing you won…
What do you suppose the damages would be for a ToS violation? The difference between subscription rates and API rates? OpenAI accused Deepseek of misappropriating trade secrets which could have serious penalties but seems like an awfully hard case to make. Seems like we’d all be better off with a law that governs data sharing among AI companies, if that’s the policy goal.
Infinite $ since you’re trying to steal their proprietary information, in their view.
Also API usage also has ToS.
Earlier quoted context omitted.
The UN isn’t a magical oracle of “Truth” - any more than an LLM is. It’s a political body, with easily provable biases, and topics you can’t trust it on.
The United Nations is biased when it found it was a genocide in August 2025: https://www.un.org/unispal/document/report-of-coi-14aug25/ The ICJ was biased in January 2024 when it said Israel's actions are consistent with genocide: https://www.icj-cij.org/node/203447 The ICC was biased when it issued arrest warrants for Netanyahu and Gallant, charging them as war criminals in November 2024: https://news.un.org/en/stor…
Earlier quoted context omitted.
The United Nations is biased when it found it was a genocide in August 2025: https://www.un.org/unispal/document/report-of-coi-14aug25/ The ICJ was biased in January 2024 when it said Israel's actions are consistent with genocide: https://www.icj-cij.org/node/203447 The ICC was biased when it issued arrest warrants for Netanyahu and Gallant, charging them as war criminals in November 2024: https://news.un.org/en/stor…
Every Western LLM will happily list you these sources, (which is likely how you got it), and would also happily discuss with you - if you’re interested - the reasons some of them may be unreliable and why your accompanying text is sometimes misleading or misrepresenting what the links actually say.
Okay please. Tell me how I misrepresented the links. I would love for you to actually do the work of reading them. Something you should've done for the past few years as they were coming out. Better late than never
Earlier quoted context omitted.
People absolutely use LLMs to research politically sensitive topics. I wish they wouldn’t, but people use LLMs as their general search engines now.
> People absolutely use LLMs to research politically sensitive topics. Note I used the word "seriously", I meant it in its fullest form, i.e. serious people. I'm not interested in what-if arguments based on "you can't fix stupid". Stupid people also blindly believe whatever a US LLM tells them without any form of verification, hallucinations and all. Most people on this planet would agree that a Chinese LLM is perfec…
This is a useless rubric and makes discussing this pointless.
> [Anthropic/OpenAI] are serving models at a particular capability level for months before their competitors, and are simultaneously applying the best models to optimizing those costs. Second, intelligence isn’t in fact a perfect commodity, in part because applied intelligence makes itself smarter Is he casually assuming a singularity has already happened? A regular first-mover advantage I can understand, but those h…
No, he was just talking about reinforcement learning, etc.
Earlier quoted context omitted.
The model is conditioned / pretrained to use a particular version of a library. The model doesn’t know why — it was just taught to use that version. To the model, it is just some insignificant detail in the grand scheme of things. It’s just like how a model would intuitively favour using English without explicit instructions — it’s not trying to sabotage other cultures, it’s just what it does.
But that would happen with literally any model without grounding and is more of a quality/competence issue, not what's being discussed. It'd be a bit of stretch to conclude open models are no more auditable than closed models based on that possibility alone. Also, in that case, there would likely be activations indicating that it is favoring a specific version. If that's an insecure version, sure that'd be suspicious…
> Maybe you can illustrate a realistic scenario in which that would be a problem, otherwise I don't really understand what your point is in this context.
I doubt it. Can you definitively prove that you can reliably detect the kind of threat I described when model weights are released? Can you be sure that your detector won't miss *any* such sleeper attacks? If not, then that's a threat that will be used to justify the ban of models (open or not) that is not sanctioned by the US government. A model being open doesn't make a difference here.
Earlier quoted context omitted.
> "Whataboutism" or "whataboutery" (as in, "but what about X?") refers to the propaganda strategy of responding to an accusation with a counter-accusation instead of offering an explanation or defense against the original accusation. https://en.wikipedia.org/wiki/Whataboutism This is a counter accusation and not a defense, ergo, whataboutism.
I never understood how people referred to or understood “whataboutsim”. I personally understand it not as a diversion but as a critique of the moral higher ground implied by the first accusation. Said differently: keep your own house in order.
Actions can be judged on their own merits regardless of who presents the arguments.
Earlier quoted context omitted.
> "Whataboutism" or "whataboutery" (as in, "but what about X?") refers to the propaganda strategy of responding to an accusation with a counter-accusation instead of offering an explanation or defense against the original accusation. https://en.wikipedia.org/wiki/Whataboutism This is a counter accusation and not a defense, ergo, whataboutism.
No, it would be whataboutism to say something like "but in America you extrajudicially kill citizens for protesting the government." It's not whataboutism to talk about the same subject in the context of a different government. In this case LLM corporations. All you're doing is deflecting valid criticism. It's not whataboutism to talk about how bad American corporations are to American citizens in the context of talk…