The ai art is atrocious - the styles are all over the place (flat cell shaded vs ghiblified vs hyperdetailed with too much text), repetition ad nauseam, varying color schemes and shapes around the same subjects (is the scaly monster pink? Green? Rainbow colored scales?), the scale is wildly inconsistent, absurd composition (the dog is going nowhere instead of the door), the ever present piss filter and just way too many of the god damn pictures.
Truly an ai post of all time - a thing that demonstrably had 0 effort put into it.
This is an article from 2024, when open weights models like llama were only beginning to emerge. With those you basically cannot reliably do any detection (as the authors admit by the end).
Which is really boiling down to text having statistically very similar properties to human generated one. Introduce a more motivated attacker and the text would be indistinguishable from real (with occasional typos, no use of "delve", "it's not x its y", emdashes and so on).
It really is a lost battle: you cannot embed extra information in the text that will survive even basic postprocessing (in contrast to, say, steganography)
3d shape classification is different from matching a set of well-known, mostly fixed patterns (like eurion constellation) necessary to detect currency.
With 3d shapes of non-governmental origin this is at best difficult and at worst intractable. Consider the fact that many parts of a gun can be split into multiple printable pieces to be later assembled, making it very nontrivial to decipher the role of the shape.
With currency, the government has the controls for the supply of the target shape (it can encode hidden signals onto banknotes) and effectively controls the relroduction side (through the pressure on printer manufacturers). But it cannot control the supply of gun-part-shapes (it is not the only source for it), and since the problem is likely intractable - neither can it enforce the control on the 3d printing side.
Paper money being almost non-fungible is a great achievement, but is it as easy to make any mesh nonfungible as well?
An essay on my experience as a dev for the past 8 years, covering the various career changes I undertook while trying to understand what I want to do and the lessons I've learned in the process.
It's a pattern among tech folks to try to solve things with technology.
It is hard to stop yourself from treating every minor inconvenience as nail for which you have a handy hammer, and I find myself overcomplicating things in my life as a result.
The goals are noble but the methods bring a lot of the complexity simply repackaged (and potentially amplified).
Cool idea! At some point I was musing about making or buying a dashboard tool like in the post, but over the years I found that I dont actually need the complexity that comes with it.
An analogue communication medium for myself and others is indeed something that might be much more impactful and human-cetric than a smart system.
Have yet to read the whole thing, but skimming the outline and dipping my toes here and there made me believe this is indeed a valuable piece of knowledge - not some absurd "how to survive zombie apocalypse" but more like "how to be reasonably prepared for mundane disasters and try to avoid most of them in the first place". Guess I found my next read! Thanks
I get that an llm can be a useful search tool over internet, but stories? Why? If someone is going to spend 5+ hours on the contents it better be made by a human.
You are actively doing a disservice to humanity by eroding trust in writing
The author describes an approach he used to manufacture a copper alloy ring that acts as an effective inductive dampener for the compass needle - similarly to the way it is done in e.g. cammengas or other military compasses - and yet allows for the use of a clear baseplate for simpler application on top of the map (without having a liquid dampener). The specifics of the alloy are not disclosed, other than its properties and general conposition, but the concept is released under CC0-like license and the author invites the manufacturers to make this a viable design to supplant existing ones.
The stakes are higher than when it comes to any previous controversial technology/bikeshed. This is not "react bad", but rather "push for trusting a black box with engineering bad".
Truly an ai post of all time - a thing that demonstrably had 0 effort put into it.