Fable 5
-
I was a late adopter of AI. It is remarkable how fast it's moving. Totally predicted months/years ago, but wild to see. The other night I built a prototype of a website I want to use to provide interaction with various products/prototypes my company is building, and I did it in like 10 minutes. Would've taken weeks, before.
-
The latest frontier model from Anthropic. Clearly the leading model right now.
But here’s the interesting part - parts of it are too risky to be released to the general public, says Anthropic. For now no one except trusted partners have access to the complete model, which includes biology, chemistry, cybersecurity, and model distillation. Everyone one else gets a version that reverts to Opus 4.8 for any question at all in those four areas. Apparently it’s too fucking good at telling you how to make bioweapons etc.
On the other hand, lots of other AI developers are criticizing them, calling it hype, asking who elected them as the arbiters of AI safety, etc. But at the end of the day Anthropic has the top models and they don’t.
Anyway, it’s probably the beginning of a new era in AI.
-
I wonder whether the differences in tool capabilities between the competitors will increase or decrease.
Usually one would expect that the competition catches up at some point, but if the "LLM improves itself" thing becomes real, a small competitive edge may soon become huge, similarly to how a small difference in interest rates can lead to huge differences in outcome after a while.
-
Anyway, it’s probably the beginning of a new era in AI.
How long do eras typically last nowadays?
I mean, yesterday the most advanced war tactic was blowing up pagers. Today it's a step-by-step recipe on how to annihilate the human race. Like & subscribe!
-
Maybe next week AI will figure out a way for me to find that thing, the name of which I can't remember, but that I put down somewhere. I'm sure I've seen it about - surely you know what I'm talking about?
-
The latest frontier model from Anthropic. Clearly the leading model right now.
But here’s the interesting part - parts of it are too risky to be released to the general public, says Anthropic. For now no one except trusted partners have access to the complete model, which includes biology, chemistry, cybersecurity, and model distillation. Everyone one else gets a version that reverts to Opus 4.8 for any question at all in those four areas. Apparently it’s too fucking good at telling you how to make bioweapons etc.
On the other hand, lots of other AI developers are criticizing them, calling it hype, asking who elected them as the arbiters of AI safety, etc. But at the end of the day Anthropic has the top models and they don’t.
Anyway, it’s probably the beginning of a new era in AI.
-
Someone prompted ‘So the mitochondria is the powerhouse of the cell, right?’ and it halted.
I assume they’ll improve this at some point and refine what gets flagged.
Hello! It looks like you're interested in this conversation, but you don't have an account yet.
Getting fed up of having to scroll through the same posts each visit? When you register for an account, you'll always come back to exactly where you were before, and choose to be notified of new replies (either via email, or push notification). You'll also be able to save bookmarks and upvote posts to show your appreciation to other community members.
With your input, this post could be even better 💗
Register Login