msdz

3 ms·

1.
by msdz 7d ago
My guess is that if you go with something other than readable text as the “thought layer”, observability becomes impossible. Which is not necessarily something the humans training a model would want re/ alignment.
2.
by msdz 12d ago
Definitely under 18 months. If I’m not fully mistaken, I checked for the combination of iPhone as main device/Android tablet as iPad-like second device sometime in the past six months, at most since the beginning of the year, and it wa
3.
by msdz 14d ago
I and not necessarily many, but at least some people I know donate (and, somewhat irregularly, will continue to do so) small-ish amounts to the foundation. [1] You’d imagine that ought to be sufficient to cover running costs. However, it
4.
by msdz 14d ago
I’m guessing it doesn’t “cancel out” due to interest.
5.
by msdz 14d ago
> But I learned something really basic Same for me! Also, just imagine being the group at Apple responsible for designing this section of the chip, starting probably almost a decade back – under the constant uncertainty of not knowing wh
6.
by msdz 16d ago
> You start really locked down, and you read and approve every request for access. Or you start out with almost every permission granted always, but the entire thing runs in a sandbox with the minimum amount of credentials needed to do i
7.
by msdz 17d ago
This is the correct response, but it also feels as though ≈nobody is sandboxing their agents/harnesses in practice. Or at least, a good majority isn’t.
8.
by msdz 18d ago
Same. Models that are fast (and thus by extension, in some direction, efficient) but not local/self-hosted IMO fill a niche for quick and cheap (and acceptable quality, of course) inference in business contexts.
9.
by msdz 18d ago
In fact, I’d go as far as to say there’s few companies I’d entrust less the level of data required for a tool like a “personal AI agent”…
10.
by msdz 20d ago
(As a Bavarian,) I’m acutely aware of the state’s economy – but my point is that SpaceX has an even bigger economy behind it, if the two were to actually be compared or compete. Good point re/ depending on auto makers, though. That als
11.
by msdz 20d ago
Agreed, but do you think either of the two federal/state governments would even have enough support material available, whether in the form of capital or otherwise (i.e. contracts), to give to Isar Aerospace that would enable them to c
12.
by msdz 20d ago
Hm, true that. I’m guessing because the typical factory was already larger/taller than human height? (Thinking of e.g. car production)
13.
by msdz 20d ago
Exactly. Existing paths, entryways, factories, etc. are human-sized and -shaped (or car, or other vehicle). Only once you build greenfield, e.g. new factories, making specially shaped and sized robots for automation jobs becomes worth consi
14.
by msdz 21d ago
> we can't block models from scraping the public repos FYI, there is the nuke option, which is generating endless nonsense pages as a form of “bot sink” [1] that ends the scraping relatively quickly, but IIRC that also tanks your
15.
by msdz 22d ago
> means... you can run for 10 seconds every minute? It’s one order of magnitude less TPS, but still, that’s the limit with just one user…
16.
by msdz 1mo ago
My take is that the difference is simply down to visibility, and by extension broader understanding, of the issue: - Cameras film me and my family and neighbors everywhere in daily life => Obviously an issue - Something about secret agen
17.
by msdz 1mo ago
> 99% of the energy going into the distro While not commenting on the rest, I just want to point out that Omarchy is not a distro (yet?). Source: https://xn--gckvb8fzb.com/a-word-on-omarchy/?:~:text=Wikiped...
18.
by msdz 1mo ago
Are you sure that’s not at least in part due to the US being the global currency and business hub, importing a good portion of other places’ wealth? Although, yes, at times unhealthy amounts of productivity probably also play a role, true t
19.
by msdz 1mo ago
And imagine plugging a small iPad screen into the thing with sidecar… Almost like a Macbook!
20.
by msdz 1mo ago
Yeah, I agree (with the same unfortunate habit) – but up to twice a day of forced work full stop is overdoing it, too.
21.
by msdz 1mo ago
So I can somewhat speak to this, as I actually did this for a while c. 2021 and 2022. The issue (apart from my Mac mini still having the old, bigger form factor back then) is that this requires a full shutdown (obviously), and that is mor
22.
by msdz 1mo ago
Have you tried eating out while not in a rush?
23.
by msdz 1mo ago
Tinfoil hat time: They saw everyone referring to Mythos, and later Fable, as the new “good” models when Anthropic released those, distinguishable from the “regular” Claude (or other companies’ models) for everyone, and didn’t have that dist
24.
by msdz 1mo ago
> Whatever the merits of MacOS, and there are many, it is not the paradigm there that all your system settings and config are stored in text files. Maybe not, but the agent can call `defaults write` too, no?
25.
by msdz 1mo ago
Hm… Are meaningful numbers of non-developers aware of, let alone using coding agents yet, which could alleviate their pain points in using an “unknown” OS?
26.
by msdz 1mo ago
Are timing (over network) attacks, physical access, etc. typically excluded from research like this for being “out of scope”, so to speak? I’m not familiar.
27.
by msdz 1mo ago
Okay, so what I’m taking away and remembering here is that if I wrote software for programmers/“power users”, I should dogfood as much as possible, right? :-)
28.
by msdz 1mo ago
> When I use LLMs I leave things as open as possible. I just give them the information they need and my ask. How do you handle security? Both “internally” against e.g. data loss, I’m assuming via limiting the harness, and “externally”, i
29.
by msdz 1mo ago
> Sure, never mind Where did you take the “never mind” part from their statement?
30.
by msdz 1mo ago
Hey, I remember reading along before and around the time you were “officially” hired – happy Kagi could recognize your work, and seemingly continues to do so :)

More ›