Hacker Newsnew | past | comments | ask | show | jobs | submit | bluebands's commentslogin

the gap is bigger than you think and widens on niche tasks


there is not much evidence there is a plateau

I hope it will plateau


I would say AI is more like hardware than software in economics and open hardware rarely wins

why more like hardware? takes very long time to develop, expensive to manufacture (run), valuable.


> As opposed to the big American shops which focuses solely on performance, they have operational costs and other factors in their constraints and requirements

This is not true, OpenAI and Anthropic compete strongly at the lower end. See Haiku 5.5 and Luna



How are they not dangerous?

There is danger with everything. I can take a kitchen knife and stab my kid.

I WILL NOT DO THAT!

The key issue is how we manage a given danger. Anthropic wants to manage the danger of LLM's in a way that protects their business model. It's in their right to do so. And it is in other's people right to recognize this approach as self serving and object to it.


Only government-approved safe people should have access to technology like the printing press or fire.

private benchmarks are good

Not really

it's not, it's two things 1. people getting used to a specific level of capability 2. people exploring capabilities and naturally finding weak spots over time

3. people vibe coded super bloated codebases, which current AI has hard time to comprehend.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: