alfalfasprout

5 ms·

1.
by alfalfasprout 9d ago
there are many things to blame POTUS for. This isn't one of them... most major fabs were already outside the US to begin with and the US chipmaking industry (well, really Intel) was really struggling even years ago.
2.
by alfalfasprout 10d ago
How is this not a frontier model? It's bleeding edge in its own niche. It's not a frontier LLM; however, applicable to many of the things people use LLMs for.
3.
by alfalfasprout 13d ago
very HN take. Reality is 3.8 flash is good at what it does (being fast + cheap)... but that market is becoming very saturated and google is not competing at all at the frontier where major spend lies.
4.
by alfalfasprout 1mo ago
Yeah but politicians get to claim this is a win. Let's be honest, that's the real reason the prosecution went for a deal this small.
5.
by alfalfasprout 1mo ago
Probably Finnegan's Wake
6.
by alfalfasprout 1mo ago
btw it's colombia not columbia.
7.
by alfalfasprout 1mo ago
If anything, AI tends to vastly overengineer things (in a bad way) which makes bloat inevitable. But I'm not sure layers of management are to blame. If anything, good management incentivize the right things and focus a team on work tha
8.
by alfalfasprout 1mo ago
Neither of them are "almost there"...
9.
by alfalfasprout 2mo ago
The weights are open, the training data obviously isn't. But a key thing you can do with open models is fine tune them. So while they may not be SOTA at everything, they can become SOTA at your particular business use case.
10.
by alfalfasprout 2mo ago
There's a massive difference between your kids vibe coding something and an engineer using AI to implement something. If you're unable to discern the difference, that's something to reflect on :)
11.
by alfalfasprout 2mo ago
On the flipside, Apple has realized AI models are quickly becoming a commodity. By not blowing hundreds of billions in speculative capex, they get to focus on the bits that likely have higher ROI: the touchpoints with AI and how it's u
12.
by alfalfasprout 2mo ago
If anything, it's becoming more true. In the last month you already have Kimi K3 reaching competitive frontier performance (in practice, a bit behind Fable/Sol but close enough). Sure, it's a chinese model but the reality is
13.
by alfalfasprout 2mo ago
Sundar Pichai began the downfall of Google by most accounts. It went from an engineer driven culture to one of infighting, politics, and bureaucracy. It's telling when xooglers nowadays complain of maintenance and improvement of produc
14.
by alfalfasprout 2mo ago
Did you even read the post? The idea is to tie conversations to code. This is a more sophisticated version of something like doltdb, not a git replacement.
15.
by alfalfasprout 3mo ago
This is going to kill someone... you can't just make these calls for people. And it's DEFINITELY not a replacement for snowpack tests, etc. at the location you're going to visit.
16.
by alfalfasprout 3mo ago
small downsides like security holes? Those aren't small. Neither is creating a codebase that's an inextensible mess that even LLMs begin to struggle with. The reality is making good decisions and thinking about approaches take t
17.
by alfalfasprout 4mo ago
This is probably one of the more level headed takes in the comment thread. There's been a concerted marketing push to frame AI maximalism as an inevitability. More or less a "it's going happen anyways so let's go all in&
18.
by alfalfasprout 4mo ago
A part of the puzzle that rarely gets discussed is something that predated LLMs entirely-- "software engineering" and "programming" have been conflated for a long time now and there's a huge gamut of roles out there
19.
by alfalfasprout 4mo ago
It's not entirely clear to me that the opposing argument is well-formed either. You constantly see numbers and statistics being wildly mis-used or overextrapolated.
20.
by alfalfasprout 4mo ago
It's not just that there's a circular deal it's that they're prevalent. And worse, with frontier labs IPOing seeking astronomical valuations that means a lot of the public is now exposed too (even if they don't all
21.
by alfalfasprout 4mo ago
> the big providers are charging full freight for inference They're not and it's not clear why you seem to believe that. The immense capex for buildouts, training costs, etc. are not rolled into inference costs. Moreover, compa
22.
by alfalfasprout 4mo ago
Generally, I agree because what happens is the messaging around AI is doing more, faster. Not using AI to deliver at a higher quality level, etc. But I think it boils down to incentives and discipline. So given the incentives we have today
23.
by alfalfasprout 4mo ago
Absolutely! Yes. This rhetoric of inevitability only benefits these AI companies.
24.
by alfalfasprout 4mo ago
Right. Which means tokens are actually being priced well under cost once you factor in all this datacenter/GPU capex. Also worth noting the datacenters are not purely for training. They're for inference too.
25.
by alfalfasprout 4mo ago
I'm actually currently studying this :) Honestly... not that dramatically. Each release is much more marginal. And quoted official benchmarks doesn't translate very well into the real world. 4.7 regressed hard in some ways. But
26.
by alfalfasprout 4mo ago
The interior is also, frankly, very meh.
27.
by alfalfasprout 5mo ago
Having tried something similar, the perceived speedup does not, in the steady state, last. To get a quality, lasting, result you're ultimately having to carefully study everything otherwise you end up quickly accumulating cognitive deb
28.
by alfalfasprout 5mo ago
Heck, you could do a decent amount with the CAS back in the TI-89.
29.
by alfalfasprout 5mo ago
on claude using bedrock it simply refuses to acknoweldge the existence of OpenClaw (Opus 4.7)
30.
by alfalfasprout 5mo ago
Fascinating, because I've seen the exact opposite across the PNW.

More ›