Is this "system card" equivalent to the stone tablets handed down to Moses? Why don't you call it "user manual"?
Do people chant the "system manual" at Anthropic Tupperware parties? Do they intone a mantra invoking Amodei's name?
Is this "system card" equivalent to the stone tablets handed down to Moses? Why don't you call it "user manual"?
Do people chant the "system manual" at Anthropic Tupperware parties? Do they intone a mantra invoking Amodei's name?
Because it's not a user manual? The idea of a model card originated in 2018 (see https://arxiv.org/abs/1810.03993) as a summary of important facts about a model. At the time, this was typically an image classifier or tabular ML model. Model cards became an important concept in AI governance, and they started expanding once models started getting more capable. The point of a model/system card is to document where the model came from and the evaluations that have been run, make a case that the model will be safe and reliable in its intended applications, and warn about any potential dangers from misuse. It's not an explanation of how to use the model.
OpenAI also releases system cards; here's GPT-5.5's: https://deploymentsafety.openai.com/gpt-5-5/safety
It used to be a "card", as in a single page or two. It doesn't make sense that they still call it that.
If calling somebody with phone is still 'dialing' someone even there is nothing round on smart phone. Then why not?
A system "card" made mostly by the model itself.
The trailing snark at the end will likely get you downvoted but I'm latching on: wtf is "system card". My previous coworkers popped that in the general slack channel when Mythos first "dropped" - "have you seen the system card" without any context whatsoever. The nerds get their clique!
Also research preview pops across new upstarts in place of beta. It's eye-rolling coming from a lifelong curmudgeon.
Just talk normal!
I'd call it a "whitepaper".
But most hype-dependent projects need new vocabulary for old concepts to keep people from looking too closely and maybe drawing parallels to "legacy" "unsexy" projects, so whitepapers get called "system cards" and startups get called "labs", and so on.
Couldn't someone else equally well argue that "whitepaper" and "startup" are hyped-up vocabulary for "report" and "unprofitable company"? It kinda seems to me like the cause and effect are in the other direction, and the vocabulary of a particular niche becomes cool and hype-sounding when that niche starts to pull in a lot of money.
yes at some point language evolves as the new normal, as designed.
My curmudgeon gripe with system card and research preview is really the parroting; so cant blame anthropic for what others do. It’s just… no, prediction markets for dogs doesn’t have a research preview.