rad_val

5 ms·

1.
by rad_val 29d ago
The weird thing is how most of the comments here are from dads saying it's hard but worth it in various ways. I've been scrolling for a while and failed to find an alternative point of view.
2.
by rad_val 2mo ago
After Opus 4.8 intelligence really started to matter less and less for the programming tasks I have. If I have to handheld anyway, why would I wait more or pay more?
3.
by rad_val 3mo ago
kinoto.io/orrery We're releasing this next week.
4.
by rad_val 3mo ago
I wasn't assuming anything. Generally speaking. The flow you describe in that comment is rather simple in my opinion and with the right harness even Sonnet would drive most of that. I judge by the ability to bugfix complex codebases a
5.
by rad_val 3mo ago
Step 1: don't trust benchmarks you don't understand - they might measure irrelevant things Step 2: test it on things you know Opus failed My day-to-day take, for the coding I do (not security related): incremental, modest improvem
6.
by rad_val 4mo ago
I'm more interested why you think my understanding is flawed honestly. I thought I distilled it decently well in two sentences. The bottom line is, in this hyperdimensional space you can find relationships that are not easily distingui
7.
by rad_val 4mo ago
The strongest argument for this is structural: what LLMs are. In a brutal simplistic way: each token is represented in a high dimensional vector. LLMs operate on them. They are the true, underlying meaning of the token for the LLM. Think of
8.
by rad_val 4mo ago
AI (in this form) will never be able to solve things we truly cannot solve yet. It might catch things that we didn't project properly or brute force things no human can , but it will never unify general relativity with quantum mechanic
9.
by rad_val 4mo ago
i haven't read their memo, but, the article talks about math being something deeply human and the AI taint. I think it's a bit of both.
10.
by rad_val 4mo ago
Agreed. As someone who was always curious but had difficulties learning math the way it's taught at the university, AI teaching me the way no professor ever could is a blessing. I fail to see the point of the memo besides: we got here
11.
Anthropic and the caravel problem(radval.me)
2 points by rad_val 4mo ago|1 comments
12.
by rad_val 4mo ago
All of them do if you don't do something about it(e.g. migrate to self hosted solutions), trusting a ToS in 2026 is as naive as it gets.
13.
Show HN: Pent – A sandbox for AI agents(github.com)
2 points by rad_val 7mo ago|0 comments
14.
There's no corpus large enough(swiftcraft.io)
2 points by rad_val 8mo ago|0 comments