4 ms·
Genuine question, why have you chosen to phrase this scraping and distillation as an attack? I'm imagining you're doing it because that's how Anthropic prefers
by jdiff 6mo ago
Genuine question, why have you chosen to phrase this scraping and distillation as an attack? I'm imagining you're doing it because that's how Anthropic prefers to frame it, but isn't scraping and distillation, with some minor shuffling of semantics, exactly what Anthropic and co did to obtain their own position? And would it be valid to interpret that as an attack as well?
- irthomasthomas 6mo agoIf you ask claude in chinese it thinks its deepseek.
- DrammBA 6mo ago> I'm imagining you're doing it because that's how Anthropic prefers to frame it Correct. > would it be valid to interpret that as an attack as well? Yup.
- fragmede 6mo agoFirehosing Anthropic to exfiltrate their model seems materially different than Anthropic downloading all of the Internet to create the model in the first place to me. But maybe that's just me?
- robrenaud 6mo agoYeah, it's different. Anthropic profits when it delivers tokens. Hosting providers pay when Anthropic scrapes them.
- jdiff 6mo agoI don't see the material difference in firehosing anthropic vs anthropic firehosing random sites on the internet. As someone who runs a few of those random sites, I've had to take actions that increase my costs (and burn my time) to mitigate a new host of scrapers constantly firing at every available endpoint, even ones specifically marked as off limits.
- 59nadir 6mo agoYes, what the LLM providers did was worse and impacted people financially a whole lot more in lost compensation for works as well as operational costs that would never reach the heights they did solely because of scrapers on behalf of model providers.
- typ 6mo agoI don't think that learning from textbooks to take an exam and learning from the answers of another student taking the exam are the same. Joking aside, I also don't believe that maximum access to raw Internet data and its quantity is why some models are doing better than Google. It seems that these SoTA models gain more power from synthetic data and how they discard garbage.