Kevin_P
1yr ago
Why would I buy AI-generated imagery from Shutterstock when I could just make it myself?
Aside from the time savings that other people have mentioned, the other big advantage of Shutterstock is that it handles all of the relevant copyrights. Using AI art generation is probably OK from an IP rights perspective, but there's still a chance that a generated image will be close enough to the source material that the original artist could sue.
Using Shutterstock already insulates companies from the risk of traditional artists giving them a copied picture. I can see how people would see the same service as valuable for AI images too, at least until the legal issues get straightened out.
Kevin_P
1yr ago
Specifically Frontier (Elite 2) and Frontier: First Encounters (effectively an expansion to Frontier but sold as a separate game to try and wriggle out of a contract). The original Elite didn't have Newtonian flight.
Kevin_P
1yr ago
The easy exercise is to try and solve exam questions from n years ago. Most of the time, in most subjects, people just walk away shocked how much harder they were.
I don't know about your course in particular, but the "look how badly current students fail old exams" technique is deeply flawed. When a course is changed it almost always adds some parts while removing others.
If the previous version covered points ABCD but the new course is now ADEF then it's only natural that today's students would be confused by points B and C. It's not a fair comparison unless you also consider how well the original group would have done on E and F.
Kevin_P
1yr ago
People here often bring up legal issues with relying on IQ tests etc but I don't think it's the real problem. The legal issues basically only apply to employers in America, but other countries without those laws are still experiencing the same sort of higher education signaling spiral.
Kevin_P
1yr ago
Sorry for snapping at you but that was one of the examples I was thinking about when I wrote the post. It makes a lot of assertions but is very short on actual numbers. There's a chart but it's on a 100 year scale and far too zoomed out to read off the emissions numbers for any given year. And maybe I'm just not reading the references properly but I looked up a few of them and couldn't find anything in there that directly says how many gigatonnes of CO2 were emitted vs how many were forecast by RCP8.5.
Kevin_P
1yr ago
Limiting to things that I've known people to forget or not think of:
-
Chargers (this is the #1 forgotten item among people I know)
-
Plug adapters, if traveling somewhere with a different electrical system
-
Mouse (you CAN work on the trackpad for two weeks, but it's much less comfortable)
-
Work shoes, if traveling in comfortable shoes (somehow shoes are much easier to forget than shirts or trousers)
-
Download or print out your tickets and hotel reservations
-
Spare phone, if traveling somewhere where roaming is expensive (it's often much cheaper to buy a local sim card rather than paying roaming fees for data)
Obviously also ID, bank cards, phone, laptop, clothes etc as others have said - but those are obvious enough that you're not realistically going to forget them.
Kevin_P
1yr ago
I've read several times here that the climate forecast often described by news articles as "business as usual" (RCP8.5) is based on outdated assumptions and that actual emissions are already well below that forecast.
The explanations seem reasonable, but when I've tried to look for data to confirm / quantify it I havent had any luck.
Does anyone have a good source with numbers for forecast emissions from RCP8.5 (and possibly other forecast models) directly compared against actual emissions since the forecast was made?
Kevin_P
1yr ago
The Oxford English Dictionary is usually a good place to look. The full version has sourced examples for early usages of all different senses of a word.
Unfortunately it's a paid service but someone here might have access, most likely through a university.
Kevin_P
1yr ago
They've changed core aspects before. THAC0 was probably the defining element of pre-third-edition D&D.
Kevin_P
1yr ago
My favorite illustration of this is something called Centaur Chess.
Early chess engines would occasionally make dumb moves that were obvious to human players. Even when their overall level improved enough to beat the top human players they still often did things that skilled players could see were sub-optimal.
This meant that in the late 90s / early 00s the best "players" were human-computer teams. A chess engine would suggest moves, then a human grandmaster would make their move based on that - either playing the way the computer suggested, or substituting their own move if they saw something the computer had missed.
But as AI continued to develop the engine's suggestions kept getting better. Eventually they reached a point where any "corrections" were more likely to be the human misunderstanding what the computer was trying to do rather than a genuine mistake. Human plus computer became weaker than the computer alone, and the best tactic was to just directly play the AI's moves and resist the temptation to make improvements.
Kevin_P
1yr ago
"Only have access to the good stuff" is probably best accomplished by limiting access to GPUs at all.
This is already happening. The US government has already banned Nvidia from selling high-end chipsets to customers in China. One important point about the bans is that this not only bans the current top-end chips but also anything they develop in the future with similar capabilities - so in a few years it will cover high-end gaming cards too, and gradually extend lower down the range as time goes on.
That's currently in the geopolitics sphere, but it's easy to see it being rolled out to other customers that the people in charge don't want to have unfiltered access to modern AI tools. If the masses want powerful GPUs they can use an online service like GeForce Now or Dall-E that restricts any sort of dangerous/undesirable behavior.
My priors were strongly in the opposite direction, at least in the fanfiction space, so I checked the numbers on Archive of our Own, (the most popular fanfiction site).
They confirmed my suspicions: M/M pairings are almost twice as common as M/F (4.7 million fics vs 2.4 million). They're also read a lot more, the most popular M/M fic had 8.2 million hits vs only 2.7 million for the top M/F fic.
More options
Context Copy link