This is where I beg to differ and say yes we should. They might just be one thief, but there is enough evidence that has already been revealed to charge someone at Meta with a crime.
How long do think it would take if you or I proudly proclaimed we had violated the copyright of millions of protected works so we could make a buck before we were ‘the defendant’? Heck, look at the poor schmucks who have been sued into oblivion by a major production company and ruined by the DMCA because they downloaded a movie, or one Metallica song!
I think we need to make an example of these fuckers. Start with Zuck and then kick over the rest of the dominoes and clean out the rest of the den.
Edit: I’m not angry at you Eric so apologies because it could read like that. It is because what you have said is true… And that makes me ■■■■■■■ angry.
Totally agree - problem is they have so much money that doesn’t seem likely. These Transformer based models - despite how limited they currently are have the potential to make people who already have more money than they can spend in 100 lifetimes even richer and the rest of us a lot poorer - so totally in line with their preferred agenda. The “We do whatever we like and two fingers to the rest of the population” is the way I see it going.
When have powerful bad actors EVER suffered for bad acts? With a few exceptions I submit never. And when penalties do come, it isn’t from justice, it is from daring to challenge a more powerful player. I submit that sanity requires that we hold this as an immutable truth. The tech empire is too powerful to control. And we should stop wasting time fantasizing about doing so. Let’s say Zuck and Co. face the wrath of government action. What will replace them? A moral actor with the needs of a safe, free society guiding his every decision?
Not. A. Chance.
The void will be filled by something even worse.
Speaking of morality, I wonder how AI’s creative appropriation handles porn. Porn is unique in that even I, a real human, can’t really understand why certain acts, physiques or imagery are compelling. Most porn is revolting even to fans of porn. Each consumer has a very narrow range of aspects that he/she finds titillating. I can see how AI engines can sort through millions of hours of anime and recognize Miyazaki’s style to a fidelity that might satisfy a casual fan. But porn? I can’t at all imagine it done successfully. I don’t want to imagine it. (Yet here I am.) Every second will have tiny mistakes or oddities that will repulse. When it gets to the point where it doesn’t repulse; where it isn’t sick comedy, THAT will be when they’ve really reached general artificial intelligence.
Unfortunately I can’t argue with that… yet these self-same powerful amoral arseholes wonder why so many ‘average’ people have sympathy for the Luigi Mangione’s of the world?
WRT porn. It is already happening - In the past few months there have been cases here where high school boys have used it to create explicit images of their female classmates or teachers. Some of them, because they created images that depict victims who are under 18 are potentially facing ‘Possess Child Exploitation Material’ charges. If they are convicted of that it gets them on to the sex offenders register for life… But some of those boys were students at the most exclusive private schools in the country and after the initial reporting; nothing, nada, zilch!
Crazy. It’s unseemly to press the point but I should have said that I was musing about “live action” video, not stills. That to me is what will make this technology clearly intelligent. Anyway, the bar tender at Mos Eisley gets the last laugh. All this time we thought he was a bigot. Turns out, a robot probably took his real job and destroyed his life.
I have to admit that I’m quite fond of using chatGPT for my mod announcements. I write a first version, prompt the AI to modestly rewrite it and remove the markup. Saves me quite some time.
Here a practical example for upcoming Vietnam War Vessels 1.0.0:
Original version by me:
Vietnam War Vessels update 1.0.0: New ships and some improvements
We decided to version this release as 1.0.0 for the simple fact that there is a breaking change in the CVA-31/Generic Essex class lua structure: We had to rename the GT.Name attribute to accomodate the sounds for the AA guns. Unfortunately this means that any mission referring to a 0.x.y CVA-31/Generic Essex will not work out of the box. But the AA guns are working now, as well as cat and island being overhauled.
On the new ships that the mod presents: A new capital ship, the infamous BB-62 New Jersey joins the Vietnam War Vessels fleet. Accompanying it are two new destroyers, USS Bigelow and USS Forrest Sherman. Some of the existing destroyers in the mod now sport an updated wake file, making the bow wake more realistic (Sumner, Laffey, Maddox and The Sullivans).
The MiG-21MF saw a fix for the landing gear: It now fully retracts into the hull and no longer peeks out. Also the nose gear suspension was animated, though the job is only halway finished currently. Behind the scenes the Liveries for the HH-2 Seasprite and MiG-17MF “Fresco-C” were rearranged, hopefully reducing disk space. Another fix was the RA-5C distance lods - previously they were sitting on disk, but were never called. Further the A-1 Skyraider landing speed was raised considerably, leading hopefully to a more realistic pattern behaviour.
chatGPT modification:
Vietnam War Vessels Update 1.0.0: New Ships & Key Improvements
We’ve designated this release as 1.0.0 due to a breaking change in the CVA-31/Generic Essex class Lua structure. The GT.Name attribute was renamed to accommodate AA gun sounds, meaning missions using older versions (0.x.y) won’t be compatible out of the box. However, AA guns are now functional, and both the catapult and island have been overhauled.
New Additions
BB-62 New Jersey joins the fleet as a new capital ship.
USS Bigelow & USS Forrest Sherman arrive as new destroyers.
Several existing destroyers (Sumner, Laffey, Maddox, The Sullivans) now feature an improved bow wake for greater realism.
Fixes & Enhancements
MiG-21MF landing gear now fully retracts, and the nose gear suspension has been partially animated.
HH-2 Seasprite & MiG-17MF “Fresco-C” liveries were reorganized to optimize disk space.
RA-5C distance LODs are now properly utilized.
A-1 Skyraider landing speed increased for more realistic approach patterns.
I’ve been using copilot AI to have a play with some brewing recipes.
For those not as nerdy about beer, creating a beer recipe is about balancing base malt volume, with flavouring malt (roasted) volume (called The Grain Bill), countering the sweetness with different bittering and aroma hops.
I like to try and recreate beer I like to drink, understandably a lot of breweries don’t publish their exact recipe/grain bill, though with various apps you can tweak the styles to get close.
So, having never tasted @chipwich brewery - Pawley’s American Pale Ale, as an experiment, I asked AI to create me a recipe, which it then tweaked with additional prompts and has created the following recipe for 57 litres of fermentable liquid.
Refined Pawleys Inspired Pale Ale Recipe (57L)
Grain Bill
85% Pale Malt (American 2-Row or Maris Otter) – 9.2 kg
8% Munich Malt – 850 g(adds a mild toasted sweetness)
7% Light Crystal Malt (40L) – 650 g(increases caramel notes for extra body)
Hops
Cascade (Bittering) – 45g @ 60 min (adjusting for a smoother bitterness)
Tbh: I know nothing about brewing so I cannot check how viable that is, but it coming from a so-called “AI” I would triple-check if it is at least plausible.
They are terribly unreliable and tend to just make stuff up because of likely word combinations. Keep in mind that they often don’t actually know what they are even talking about, even if it looks like they do.
I’ve plugged the bits into my brewing App, this is how I record and create recipes (see, I told you I’m a nerd) and it’s within scope of an American Pale Ale….
Ok, that sounds good indeed. At least it ensures that you are producing something that is beer of the right category and the right amount of it.
Can’t be too wrong then.
