fl0id

13 ms·

61.
by fl0id 1y ago
Not cynical, realistic
62.
by fl0id 1y ago
The immediate no for scientific collaborations is zotero support being paid only. If it was just one user, sure they could decide to pay. But no way I am convincing others to sign up. (If this is somehow possible without paying while not ‚s
63.
by fl0id 1y ago
I'd really like them to bury this once and for all. It's really exhausting that it's like an undead zombie that always comes back.
64.
by fl0id 1y ago
LOL no. Here right wingers just want to defund higher ed and control what they do. None of which will help with quality of output. (and those rankings are largely useless anyway) they don’t want more educated ppl, because they blame educati
65.
by fl0id 1y ago
To blame are employers but also politicians and Uni admin staff enabling the narrative that a degree should always be about employable skills. First employers often eintreten know what they need and in general this focus on practice is exac
66.
by fl0id 1y ago
There is no neutral. It will just be biased based on its training data etc.
67.
by fl0id 1y ago
Because there’s a law against it in Illinois.
68.
by fl0id 1y ago
That’s actually a example of sth different. And as it’s basically a placebo it only harms people’s wallets (mostly). That cannot be said for random llm failure modes. And whether it can be prescribed by doctors depends very much on the coun
69.
by fl0id 1y ago
When has technological progress leveled the playing field? Like never. At best it shifted it, like that a machine manufacturer got rich in addition to existing wealth. There is no reason for this to go different with AI, and it’s far from c
70.
by fl0id 1y ago
This. At least here therapists don’t have a problem getting new patients.
71.
by fl0id 1y ago
You can also write to FSFE about it, afaik they give legal advice in foss cases.
72.
by fl0id 1y ago
That’s just a general explainer of subscription models. As of right now VC money is necessary for just existing. And they can never stop training or researching. They also constantly have to buy new gpus unless there’s at some point a plate
73.
by fl0id 1y ago
at the end, it says it was at least tested to compile (and presumably run) on real machines.
74.
by fl0id 1y ago
this already existed with things like POE. like I mean for normal users, with a subscription for multiple models. same with frontends for api keys.
75.
by fl0id 1y ago
Actually even of their was AGI, it would be even more necessary to control it.
76.
by fl0id 1y ago
I call BS. That graph seems very misleading, like just getting faster for me is not improving exponentially. By improving exponentially most ppl would understand getting smarter
77.
by fl0id 1y ago
You can usually get the rights but it costs more. Like in software if you want an exclusive license…
78.
by fl0id 1y ago
Nile has even a masters in chemistry I think, and very likely at least a business entity. So might work. And even for him some things are hard to get, a Canada has similar restrictions.
79.
by fl0id 1y ago
Your first point would be more fitting if a language checker would need a complete, computable grammar that can be parsed and understood. That would be problematic for natural languages.
80.
by fl0id 1y ago
Word has the worst performance ever. So I don't think competing with Word on performance should be hard.
81.
by fl0id 1y ago
Yeah same. Super worried
82.
by fl0id 1y ago
Which is exactly what the borg suggest in their issue.
83.
by fl0id 1y ago
For the model it’s not. But I think many of these bots are also from tool usage or ‚research‘ or whatever they call it these days. And for that it doesnatter
84.
by fl0id 1y ago
AFAIK it will have f e access to your account, browser info and f e location information. Just from that it can figure stuff out. Some guy tested that when they asked to locate a photo.
85.
by fl0id 1y ago
Same. Like Claude code for example will write some tests. But what they are testing is often incorrect
86.
by fl0id 2y ago
I count 3 lines. And that is only if everything works and you have conda etc installed already
87.
by fl0id 2y ago
What did you train it on? Would be great to k ow if I’m opening myself up to liabilities by using it
88.
by fl0id 2y ago
But how would you confirm it if there’s no ‚reproducible build‘ and you don’t have the hardware to reproduce?
89.
by fl0id 2y ago
though they wouldn't need to use MLX, could also use pytorch etc
90.
by fl0id 2y ago
No. He just wants people to have some decency and standards in communication and not arbitrarily decide that their little kingdom can never be touched.

More ›