3 ms·
This is because the users are training the product. They need training data, so they sell inference at the price of power.
by genxy 5mo ago
This is because the users are training the product. They need training data, so they sell inference at the price of power.
- gruez 5mo ago? >API Services . If you use the API services, we will collect your IP address and the content (text, audio, video, picture) you submit to analyze the relevant instructions based on the model you select and to generate the returned content. Xiaomi will not use the content you provide for model training or any other purposes. https://privacy.mi.com/XiaomiMiMoPlatform/en_GB/ https://privacy.mi.com/XiaomiMiMoPlatform/en_GB/
- colechristensen 5mo agoAnd what legal recourse do you have if they don't follow those rules?
- windexh8er 5mo agoYou have no recourse in the US, either. Trust no one is the only path given all of the training data is stolen in the first place. It will come to light that one or many of the Frontier providers held the data, changed ToS and trained later minimally. But I think they just don't care and will train regardless. None of them abide by any level of ethics that would actually prevent them from leveraging an opportunity.
- koteelok 5mo agoChinese corporation would never lie
- Tiberium 5mo agoChatGPT (the setting is shared with Codex) and Claude (shared with Claude Code) also have sharing enabled by default, so why aren't they cheaper?
- koteelok 5mo agoThey are? They give away thousands of dollars via subs.
- Springtime 5mo agoThere's evidence various third-party models (including Deepseek) used distilling in training, based on models from those leading services. So they have more flexibility with pricing.
- behnamoh 5mo agoSo? And Anthropic/OpenAI literally stole copyrighted content to train their models.
- Springtime 5mo agoThe point was that distilling based on others' models for training means they're not spending the same amount on R&D and/or training, giving them headroom in other ways (responding to the parent's point). It wasn't a comment reflecting on copyright/fair use.
- behnamoh 5mo agoIn the same fashion, Anthropic/OpenAI also reduced their training cost by not purchasing the license to copyrighted work and stealing it instead.
- malnourish 5mo agoIs that fundamentally any different than what e.g., Meta and OpenAI have done? Besides, hasn't SCotUS ruled that raw LLM output isn't subject to copyright? So these companies would be breaking a ToS at worst.
- camelmel 5mo agoIs this training data even valuable? Usually AI data annotators get paid to write LLM responses, but here all they'd be getting is a bunch of user queries.
- VerTiGo_Etrex 5mo ago1. Feed the same queries into Claude 2. Train on the Claude responses 3. ??? 4. Profit This has been the strategy for months now