thepasch

6 ms·

61.
by thepasch 4mo ago
Am I crazy to be extremely suspicious about the fact that this heavily security-focused task suite didn't trigger a single of the infamously hilariously overparanoid guardrails? This, along with the fact that the model "cheate
62.
by thepasch 4mo ago
As per usual in situations like these, one must look at the actions in order to assess whether there's any worth in the words. And the actions of Anthropic have, by and large, been steering hard towards establishing a walled garden, em
63.
by thepasch 4mo ago
Working to keep a roof over the head of yourself and those you love is a necessity . It can become an identity if you enjoy what you do, sure, but that is not a given for, I'd say, a big majority of the workforce, globally.
64.
by thepasch 4mo ago
> Yeah I think there are ways to know, ways involving less dependence on a LLM. This kills the entire value prop of using LLMs as research accelerators, though.
65.
by thepasch 4mo ago
What it feels like to work with Fable: > Switched to Opus 4.8: Fable 5 has safety measures that flag messages on most cybersecurity or biology topics. They may flag safe, normal content as well. These measures let us bring you Mythos-lev
66.
by thepasch 4mo ago
Yeesh. Anthropic's paranoia about China is starting to get pathological.
67.
by thepasch 4mo ago
The sooner we all collectively realize that the commonly cited "reasons why LLM code is bad" and "reasons poorly architectured, poorly planned, poorly tested code is bad" are a circle on a Venn diagram, the sooner we can
68.
by thepasch 4mo ago
> I also don't quite get why one would want such a setup Because LLMs are trained on UNIX-style shell usage and, in general, much more competent and efficient at using that over PowerShell or Batch/CMD. They want agents to run
69.
by thepasch 4mo ago
From experience, whenever someone asks in that particular tone and is actually provided with examples, they proceed to bend over backwards to "prove" that it's secretly not much of an improvement at all/AI psychosis&#x2F
70.
by thepasch 4mo ago
That, as an LLM might say in this context, checks out!
71.
by thepasch 4mo ago
I'm honestly flabbergasted that everyone's implicitly accepting that it's "people" who wrote this blog post. This reads exactly like the distorted half-true nonsense an LLM would confabulate together from a cursory
72.
by thepasch 4mo ago
This seems like a meme tweet? Is there any indication whatsoever that this random Twitter person would know about such a thing?
73.
by thepasch 5mo ago
“Outdated” was perhaps an inaccurate term to use here. The issue isn’t what dates these studies were published on, it’s what models they were evaluating on. This field is moving faster than a speeding bullet. To start with the most recent
74.
by thepasch 5mo ago
Another way to be wrong about AI-assisted coding: basing an entire article on mostly outdated empirical findings, some of them wildly so?
75.
by thepasch 5mo ago
As long as vendors can’t pay their way into preferential treatment, this is objectively a goo– –BWAHAHAHAHAH nah couldn’t finish that sentence.
76.
by thepasch 5mo ago
If only the American justice system displayed a fraction of this same raging fervor when it came to crimes that actually caused harm to someone.
77.
by thepasch 5mo ago
Google just killed Google as we know it. Kagi, DuckDuckGo, et al. still do search as we know it just fine.
78.
by thepasch 5mo ago
> I can't read minds. I know I have one and you know you have one. That's enough for both of us to know that mind is a real phenomenon. So that's a religious argument, then. It's real because enough people believe tha
79.
by thepasch 5mo ago
I don't follow? I can copy a file and then consider the two files to be separate copies of the same data ? What should I have said instead? "We"? "Him and it"? Self-modeling is part of my experience. I'm sur
80.
by thepasch 5mo ago
> Why some kinds of information processing and not others? I have no idea. If that's what the hard problem of consciousness boils down to - we don't know why some complex math produces consciousness and other complex math doe
81.
by thepasch 5mo ago
> As for non-scientific proof that you have a mind, you can just observe that for yourself in every instant of your own personal experience. Two things here: 1) How do you know I have a mind? How do I know you have a mind? 2) What is
82.
by thepasch 5mo ago
I suppose complexity defines to me as something that's inherently hard to understand. The definition of "complex" is "difficult to understand." The move I don't agree with is proceeding to posit "because
83.
by thepasch 5mo ago
> If all information processing leads to consciousness Did you actually read what you just responded to?
84.
by thepasch 5mo ago
This does not seem like a particularly difficult question to answer to me, and I suspect it's because I'm not particularly precious about what it means to "be me." The logical answer is that this spreadsheet, supposing i
85.
by thepasch 5mo ago
Explain psychedelics, then? Do psychedelics have access to this supposed "separate layer" that mind exists on over matter? If yes, how? If not, how can something that ostensibly only interacts with the matter have any effect on
86.
by thepasch 5mo ago
This is a religious argument. If you want to go down that path, then sure; but I suspect that's not what you actually believe.
87.
by thepasch 5mo ago
That is a lot of anger to come out of what essentially boils down to "I don't believe/want to believe that complexity can cause subjective experience." And this bit: > I don't believe the "soul" needs t
88.
by thepasch 5mo ago
> Either all the computation happens "in the dark", as in a calculator or an Excel spreadsheet or a slide rule or Factorio, in which case we are p-zombies and consciousness is an illusion, which contradicts every waking moment
89.
by thepasch 5mo ago
> Harness engineering is a moat. I mean, if that’s the case, then Anthropic themselves are currently actively filling in that moat with nice, solid, walkable dirt. Claude Code may have been a moat 6 months ago but these days you’ll wan
90.
by thepasch 5mo ago
Article title: “The US is winning the AI Race” Article content: “The US are capitalizing on AI the best” A lot of assumptions there that no one can actually verify as true right now. If commercialization into rent-seeking SaaS landscapes

More ›