svachalek

6 ms·

61.
by svachalek 2mo ago
The way this article defines happiness, as a general success at ill defined problems, would make powerful people happy by definition imo.
62.
by svachalek 2mo ago
This is my personal theory for the cause of this style: Ouroboros. The official OpenAI explanation for how ChatGPT got obsessed with goblins blames it on exactly that: --- That creates a feedback loop: - Playful style is rewarded - Some rew
63.
by svachalek 2mo ago
Because it's thick.
64.
by svachalek 2mo ago
I've hardly met anyone outside the tech elite who likes it. I can think of one doctor who's a counterexample.
65.
by svachalek 2mo ago
If you're talking Opus pricing, it's more like $200/day.
66.
by svachalek 2mo ago
I'd put it on this. In my experience Python is fine for scripting your own machine but an obnoxious platform to distribute code on. It's very fragile to version changes, in both directions; I don't know how many things I&#x27
67.
by svachalek 2mo ago
Maybe it's me but I don't see how DS Flash is better than GLM at all, much less by a huge gap. I'd probably protest less against Fable and Opus being put at the same level than many would, but there's no denying the two
68.
by svachalek 2mo ago
AI tends to really like staccato writing, a series of sentences that are too short and simple. It's evident here, although broken up with really long sentences. Tortured and near-miss metaphors are another thing it likes and it's
69.
by svachalek 2mo ago
I think in the abstract there's a lot to be concerned about with that. However, I don't see it in this particular case. It's a real customer with a real order and real money, going through proper business channels to place an
70.
by svachalek 2mo ago
Unfortunately I don't think it's possible for companies not to abuse your information, unless it's like 1-2 people and they're independently wealthy and the business is really more like a personal hobby. I don't kno
71.
by svachalek 2mo ago
This is such a warped and cynical view of history. Yes, advertising existed if the binary existence is what matters to you, but most online spaces were nearly free of it until the DoubleClick days -- that's 25+ years. Even then, huge s
72.
by svachalek 2mo ago
Agreed. The combination of social media (walled gardens and plunging quality of content), advertising (I know what you were looking for but have this word from our sponsors instead), SEO (more advertising but we didn't pay for it), and
73.
by svachalek 2mo ago
These are not 24 hours of inference with floating point errors accumulating; largely the system guards against errors compounding. Tool failures, compile failures, test failures, etc, push back against the model taking a wrong turn and forc
74.
by svachalek 3mo ago
Let's take world leaders as an example. Suppose one starts making up everything they say on the spot, never researches or cares about facts, and just says things people like. Would this world leader be (a) less or (b) more trusted and
75.
by svachalek 3mo ago
It seems telling that none of the photos show anyone wearing it.
76.
by svachalek 3mo ago
It's a lot of stuff, but with the possible exception of Waymo and the TPU, isn't this all going down in history under "also ran"?
77.
by svachalek 3mo ago
Why were you buying beef before, then? Hasn't pork always been cheaper than beef?
78.
by svachalek 3mo ago
That said though, Flash isn't it. The prices on the latest flash models put Sonnet and Terra to shame.
79.
by svachalek 3mo ago
gpt-2-image has such a yellow tinge though. It stands out badly on nearly any screen. The image quality is great otherwise, but I generally prefer nano banana for its color balance.
80.
by svachalek 3mo ago
Nah, people are vibe coding lots of apps that no one cares about, but this is more like going through Stack Overflow and answering the most upvoted issues that don't have answers. These are published problems that have prior interest,
81.
by svachalek 3mo ago
We've come a long way since the first LLMs were created. They originally produced a lot of surprises that were very hard to explain, it's true. But current research is very tightly directed at producing forward results; it's
82.
by svachalek 3mo ago
That's an argument for not using the word for anything, not specific to this case.
83.
by svachalek 3mo ago
And how do glial cells and calcium waves "obviously" create a fully conscious mind while tensor operations do not?
84.
by svachalek 3mo ago
Largely hypothetical, not very effective ways to fight against it. Meanwhile we have the entire Pacific Ocean and US Military between us and China.
85.
by svachalek 3mo ago
I saw this even on the weekend when it wasn't under heavy load. We agreed a problem was due to one choice of word in a prompt and then it went off and wrote 600 lines of code to "fix" the issue.
86.
by svachalek 3mo ago
Cleaning the house doesn't retroactively fix the neglect though. It just makes the house look cared for to the lizard brain.
87.
by svachalek 3mo ago
The guy next to me at the last movie I went to was scrolling TikTok. Seriously what the hell, you can do that anytime and you actually paid to be in that seat.
88.
by svachalek 3mo ago
It would be interesting to remake it, replacing the premise that people are genetically stupid with just the idea that they're habitually stupid from generations of reliance on AI and social media.
89.
by svachalek 3mo ago
Some of this stuff is ludicrous. I finally tried /loop last week and discovered every loop iteration passes the entire context history to the model. So pretty quickly you're running a full 1M context window, without cache, likely
90.
by svachalek 3mo ago
I haven't seen either of these running outside their creator's services yet, but typically you can watch services like openrouter or nano-gpt for it to show up at a (usually small) discount.

More ›