wnkrshm

I would imagine that to propagate any confidence value through the system you'd need to have priors for the confidence of correctness for all data in your training set. (and those priors change over time)

But the social nature of art also means that humans give the originator and their influences credit - of course not the entire chain but at least the nearest neighbours of influence. While a user of a diffusion generator does not even know the influences unless specifically asked for.

Shoulders of giants as a service.

No accounting for particles yet, which you'll also keep hitting, making your ship's materials radioactive and causing lots of secondary particle showers, bremsstrahlung and the likes.

Adding to those arguments, you can include exports/imports into the CO2 emission estimation and suddenly one sees that a good part of China's CO2 emissions are for products that are exported to the EU and the US (Edit: see e.g. [0]).

So shifting high-emission production to China and then pointing the finger at China for products we consume is kinda dishonest.

[0] https://ourworldindata.org/grapher/share-co2-embedded-in-tra...

I wonder why Musk is always thrown into the ring with SpaceX, when its entire operation is mostly done by Gwynne Shotwell.

appropriation also includes a resonance and admiration of the original - you cannot admire if you don't know where you're appropriating from

I feel it's also a scaling issue: If you wanted to do all of the digital bookkeeping of today on paper, that would be perhaps impossible. I don't think personal note-taking is in any way a large contribution to paper use.

I do art on paper and even I don't go through 500 pages in a year. It's at work, things like 60 page reports that need to be printed for submission (and then get printed 2x or 3x again because you can replicate it).

The ease of printing documents is the culprit, you can just replicate large physical artifacts again and again. You can't do that with your personal notes that probably only exist as unique artifacts and maybe get scanned into a digital archive.

Regarding your other comment: why wouldn't a high-altitude detonation, even outside the atmosphere cause an EMP? I feel like the gamma photons emitted in space will eventually hit the atmosphere and with that cause electrons to spiral along field lines. Isn't the question just one of intensity?

Or is it largely dependent on multi-photon interactions to impart enough impulse on the electrons?

There are people who argue that there are ways to keep limited nuclear warfare limited. [0] I think the RAND institute also published some study on it not outright rejecting the idea but I can't find it atm.

[0] https://www.sup.org/books/title/?id=21511

It entirely depends on where you see the role of a "fighter" but I think we both already agree that it won't be anything like naval battle on Earth.

Maybe you're familiar with Atomic Rockets, the collection of engineering and science discussions about hard scifi? If not, I may have got a rabbit hole for you that will eat an entire week, easily: [0]

(Edit: while the page starts off dismissing fighters for many reasons, there are discussions of many concepts with their merits and possible defenses of the concept)

[0] https://www.projectrho.com/public_html/rocket/fighter.php

The models can produce a very very biased memory (as any other form of history) - just by virtue of being a product of technology does not make these algorithms any more objective. They're trained on biased data after all.

So what we'll remember is the cliché form of everything. That's worth something in terms of history but is it an improvement over primary source material?

There are some in-universe laws or facts that fictional genres cannot give up (Edit:typo) easily - one in scifi is the small crewed spaceship or even space fighter.

Even if guns exist, there is an arguable thrill to reading about a character infiltrating a castle with a saber and two pistols that cannot be reloaded easily. Long ago, those were pirate or naval pulp fiction.

Imo any setting can produce great stories, it's in how they're told not in how the world works. Established genres of fantasy are just attractive to some people, since analogues of the laws and structure of the world are already partially known - and it's possible to subvert expectations while not alienating people.

Isn't this the same issue as with the neuro-visual reconstruction techniques matching on a small collection of samples?

Starting small doesn't work if you have a nonlinear system. In nonlinear systems, the magnitude of a perturbance isn't linearly related to the absorbed effect.

So in the worst case: small input, giant irreversible effect.

I feel any method that is powerful enough to dependably act as an attourney is powerful enough to found its own company, write software, acquire clients, subcontract etc.

With the expression I meant the realization of a business idea into a self-sustaining business.

The irony may be that removing the barrier to legal advice in that way will at the same time make the implementation of your startup a triviality.

In physics in college we were grilled in verbal discussions with professors and an assistant for ~30 minutes or until the professor had found the limits of our knowledge, one I had went on for 50 mins.

The entire grade of the diploma was composed of all the verbal tests and the grade of our thesis (pre B.Sc.) - so there had to be a point where you knew virtually everything well enough to explain it at least in one way and derive some relations or sketch a derivation if prompted to.

The flaw here was the personality of professors - some were easy to piss off, others more patient.

Then the alias itself has to be protected though, else someone could harm you by impersonating your alias.

I didn't know the background but I'd agree though not necessarily as strongly as hating it.

To me, the poem paints an unreachable ideal of being that just doesn't resonate with me at all. It may just be me but in my mind it feels naive.

I think the entire longtermism thing is at the heart something that Camus discusses in the Myth of Sisyphos: We will ultimately all die and be forgotten, what good is struggling, working and living?

That is the absurd condition every human being finds themselves in. Camus sees two solutions that have been suggested so far in philosophy:

Suicide (which he rejects outright as too easy and lazy) and a "leap of faith". The "leap of faith" consists in finding by passion some irrational, maybe larger purpose outside of what can personally be known or experienced, a belief that negates this absurd condition. Historically, this is the domain of religion and spiritualism.

Longtermism seems to me just that: an existentialist escape from nihilism by the promise of some kind of vague purpose with vague metrics and an unknown, irrational payoff you will never know (at least for those subscribing to it without motives of profit).

I don't think that future will come to pass, because there is no application of "AI" on the horizon that gets climate change halted by 2030. And the consequences, which are already beginning, will royally dash any hope of utopia like that in the near future. Let's call it 2130.

As to the question: work is still work. People tend a garden and are happy with what they made, despite being able to shop for vegetables in a supermarket. But being in control of your own creation has its own reward and it is appreciated by others, e.g. when you personally cook for them with what you personally grew. Or when you hand-paint something for someone, etc. Success at work is only a narrow dimension of success.

Chillwave like this [0] or the original Ghost in the Shell soundtrack.

Edit: Also great to block out the environment is this [1], I donated 50€ years ago and still get to listen to everything.

[0] https://hellometeor.bandcamp.com/ [1] https://mynoise.net/

Even in a perfectly deterministic universe, the accessibility of that determinism to models hinges on knowledge of the state of the universe.

We're surrounded by highly nonlinear interactions that we just approximate as linear - if we want to truly predict anything, in general, we'd need infinite precision on the variables that describe the state of these systems, because an infinitely small perturbation in the initial conditions can cause an almost (limited by conservation laws) arbitrary deviation in the predicted state.

To be very cynical: We are trapped in a hellscape with no chance of ever breaking free, it does not matter one bit if it's simulated or not.

Sisyphus teaches the higher fidelity that negates the gods (or simulation) and raises rocks.

That hit me years ago when working conditions at Tesla came to light, I realized Elon can never have his Mars colony be self-sufficient or he'd be thrown out of an airlock.

I doubt that, looking at how bad the SOTA in detecting generated text is or conversely how human-like the generated text has become.

Maybe it's like in the free-to-play video game business: Having one whale (person who spends more on the product) outweighs having lots of smaller customers.

The IMO dystopian move is to just let the AI employ humans as 'hands' for subsistence.

how much do the weights have to change for it to be transformative? Edit: you arrive at the same question as taking an image with img2img and running it through a diffusion variant - keeping the composition, colors etc. but no the details

I have a report where the actual project work done is not exactly what was planned, so that's triggering some avoidance. It was hard to start with directly writing the report.

So what I did is the classic 'break it down' but with a pinch of artificial compartmentalization from the part triggering the avoidance:

I tried to focus only on what was actually done (significant in volume of work) switched to itemizing everything that was done first, not directly working into the report template.

I report on each of these items in isolation first, without reference to the plan, which gives me momentum since the pages fill up fast.

The last part is sorting the pages into the report template, according to the work items they fit best with.

Edit: The idea being that the momentum from the documenting of the work in isolation, will carry me through producing relatively little text referring to the plan.

It's not honorable work but it needs doing.

Yeah, I'm dealing with it at work right now for a deadline. I'm using my entire tool box.