Wir sollten uns Gedanken machen, wenn schon die Beschäftigten von Amazon vor der KI warnen, die sie selbst gebaut haben und mit der sie ihre Geschäfte machen. Wohlgemerkt, es heißt Geschäfte und nicht Gewinne, denn die allseits diskutierte KI macht (noch) keine Gewinne, sie wird nur zu einem Fetisch aufgeblasen und sie kostet uns ...
"In einem offenen Brief fordern über 1.100 Amazon-Beschäftigte ihren Arbeitgeber zu einem radikalen Kurswechsel in der KI-Entwicklung auf", schreibt die taz und weiter heißt es in dem offenen Brief: "Wir sind überzeugt, dass die überstürzte Entwicklung von KI um jeden Preis immensen Schaden an unserer Demokratie, unseren Jobs und unserer Erde anrichten wird."
Und dann nennen sie ihre Gründe:
Amazon muss seine Absicht klimaneutral zu werden wegen des Strombedarfs aufgeben,
die Technologie wird zunehmend für moralisch fragwürdige Zwecke eingesetzt,
Massenüberwachung,
autonome Waffensysteme,
Trumps Abschiebewahn, die Polizei-KI Palantir identifiziert für die ICE "illegale" Migranten (Palantir läuft auf AWS Servern),
Amazon will 14.000 Jobs streichen, weil diese demnächst von KI übernommen werden sollen,
dabei ersetzt KI keine Arbeitskraft, sondern lagert sie nur in Billiglohnländer aus, wo sie zu stupiden Trainings- oder Korrekturarbeit verkommt.
Vielleicht sollten sich auch die Kunden von Amazon dazu mal Gedanken machen ...
Mehr dazu bei https://taz.de/Amazon-Beschaeftigte-warnen-vor-KI/!6135282/
Kategorie[21]: Unsere Themen in der Presse Short-Link dieser Seite: a-fsa.de/d/3M2
Link zu dieser Seite: https://www.a-fsa.de/de/articles/9381-20251220-unmoralischer-einsatz-von-ki.html
Guess the holidays are good for finally writing up some old hunts. Badness is I can't find any elegant way to make my Batman meta work. Going to have to use a very lame shell meta since there's only 2 weeks left.
Forgot about this write up since Boston was not the last location to run, and I sat on it for a whilw.
Palantir Puzzle Hunt (Sept 29, 2012)– The League of Extraordinary Time Travellers
I stumbled across this hunt at the MIT Career Fair, I’d never heard of it before. Apparently, it’s in the second year of running and is run at several campuses across the country. I’m not sure of if it’s supposed to be for undergrads, but they didn’t seem to mind. The hunt is very much in the style of DASH. The puzzles follow a linear path, and answers are associated with a map to give you the next location to go to. Scores are done based on the time you take on each puzzle, and hints are available for a reduction in points.
It was a very well organized hunt. Each team was provided a tube of supplies including everything they would need in the hunt. Scissors, a ruler, pencils, a chessboard (much to our apprehension), and even a flashlight were in the tube. The flashlight was apparently to be used if it got late, but had no bearing on a puzzle. They also provided lunch and dinner. Actually, there was way too much food. I image there are some very happy undergrad dorms feasting on the leftovers.
The event had a loose time travel theme. Apparently, there was originally an opening skit of some sort that was not done. All of the puzzles were fairly straight forward and there was a nice variety of puzzles. Also, they were all beautifully printed on fairly high quality print stock. Most of the puzzles had flavor text that heavily hinted at the puzzle mechanics, although these were buried in a lot of story text (thought I never could quite piece together what the narrative was).
DFA fielded a team, though our 6th got sick and didn’t show. This may have slowed us down a bit, as he is perhaps our best solver, but honestly, 5 people already meant that we frequently had people that couldn’t really contribute. This might be because we’re bad a parallelizing. I’m not sure. In any case, despite a major screw-up and some minor obtuseness, we came in 3rd and got some hand towels. First place got beach towels. I’m not really sure why that was the prize. Only one puzzle had any Hitchhikers’ Guide theme that I could see. Anyway, it was fun and a very well run event. Many thanks to Palantir!
Onto the puzzles (some spoilers ahead):
Rodentia Overlords: The puzzle was very straightforward and had a loose Hitchhikers’ Guide to the Galaxy reference which didn’t seem to matter in any way. The puzzle involved many colored squares and this was where we were first impressed by the quality of the printing. The flavor text very clearly stated what to do, at least as far as I was concerned. We were able to successfully parallelize the first part with me finding the path and two guys writing down the message. The message led us to search the grid for certain squares and we actually had a really hard time searching the grid for the squares we needed. We also once again relearned the lesson that solving at the top of a bunch of stairs is stupid as dashing down them to submit an answer is a terrible idea. I think we were about the fourth team to leave, a good start. Each puzzle answer corresponded to a numbered two word phrase on the map so if the answer was “jokes” and “witty words” as number 20 on the map, you’d go to location 20. This puzzle sent us inside where it was nice and warm.
Mass Anarchy: This was a Monopoly puzzle. Each clue referenced a location on the Monopoly board by cluing words that, when put together, we homophones to a location. For example, the first one we figured out “Like Yosemite’s area” was referencing “Park Place”. The clues were paired and connecting the pair clues gave us letter to spell out our answer. Sadly, we were somehow not able to get this until we solved all the clues. For some reason we just couldn’t come up with the word without the letters which is pretty pathetic. Fortunately, this only took a few extra minutes.
Alien Invasion: Back outside. We must have been one of the first to arrive. This was the inevitable chess puzzle. We’ve never had much luck at these. We solve them, and yet somehow it always seems to take far longer than should be necessary. The puzzle was actually very light on chess knowledge. You pretty much just had to know how pieces moved. It was more of a logic puzzle as the back ranks had been scrambled into an unknown starting position. You had to figure out the initial configuration from the moves. We actually had this figured out and had played out the whole game in probably 20 minutes. However, we completely screwed up getting the answer. We were getting strings of letters that seemed to make no sense. It took us another 30 minutes or so for us to see the answer. We tried many things that made sense. And many that didn’t. There were three places were the puzzle extraction could go one of two ways. In all cases, one way was clued as correct. I’m not sure how, I wasn’t looking at the sheet (since I was the one stuck playing out the game several times), but apparently at no point did we right down the most obvious (i.e. clued at) combination of letters until the very end. This was very depressing. We’d been doing well, but now several teams had moved past us. This was the low point of the hunt.
Tectonic Slip: This was actually a pretty cool puzzle. You had to order a bunch of pictures so that they would show how the plates have shifted (and will shift in the future). We probably gained a little bit of time here as we had extra pairs of scissors. This was also really the only point in the hunt where another person would have helped, as it took us a while to arrange all the images in order. Each image was associated a letter, which spelled out a phrase explaining how to get an answer once all the images were in order. We found out later that there was a clip in the tube that was supposed to be used to make a flip book, which would make it easier to see how things shifted. This would probably have been easier than our chain which was several feet long.
Technological Singularity: This was a mirror puzzle involving sharks and lasers (at least in theme). We spent a good 4 minutes figuring out the message telling us what NOT to do. I believe James was the first to notice the unique numbers of the edges which gave us a message telling us the next step involved colored lasers. We wasted a few minutes after completing the second step wondering how to get an answer, before I pointed out the correct extraction would probably be where all three colors of lasers overlapped. Unfortunately, we had not marked our lasers clearly, with the conveniently provided colored pencils, and we had to spend some time disambiguating our lasers.
Thermonuclear War: The puzzle seemed to have an irrelevant Dr. Strangelove theme. This puzzle involved logic gates which controlled a 7-segment display. It seems I don’t recognize logic gates anymore. How embarrassing. Grant and Robin could still read them off, but the rest of us had to use the provided cheat sheet. Unfortunately, our first attempt to determine which letters were “ON”, namely using the first letter of all bombed cities, was a failure. We worked out each diagram, and inverted our choice of 1 and 0, before I realized that the mechanism we were using to determine which letters were ‘ON’ was really dumb. There were four bombed cities, and four diagrams. Each city name corresponded to one diagram which would have those letters on. Even after using the right mechanism, we had the gibberish for a while because we’d filled in the grid incorrectly. At least we remembered to check our data this time before going off on more idiotic ideas.
Pandemic: We sucked at this. This was supposed to be a simple-minesweeper variant. However, we didn’t think of this variant until the par time was almost up. I think I got us stuck on the fact that normal minesweeper didn’t work. Robin eventually figured out that there can be multiple bombs on a square. I don’t think we realized until afterwards that there was 1 bomb on the first number and 2 bombs on the second. Fortunately, our assignment of values was confirmed by the organizers. We were stuck for a bit longer until I commented that maybe the big board was also minesweeper, but that I didn’t think it would be constrained enough to give a unique solution, but thankfully Grant wasn’t as quick to write it off as I was. We got the answer a bit later after a truly dazzling display of how bad our addition skills are. It was rather sad that it took us so long to think of reapplying the mechanic. We did learn that Robin can recite a QWERTY keyboard from memory which is a little frightening.
Meteor Impact: The first part of this puzzle was an extremely unmotivated conversion of dots to letters where 1 dot is ‘A’ to 26 dots being ‘Z’. I’m really not entirely sure why this step was added in. This tells you a variant of the 5 by 5 grid that is used for the tap code. I’m not entirely sure why they didn’t use a standard grid. It seems like it wouldn’t have been a problem. In any case, the tap code on the second page gave a pangram (almost). Each letter, in addition to being in the tap code, was associated with a subset of the rainbow colors. Also, there were 7 dashes, one per color, at the bottom of the page suggesting one letter per color for the answer. We struggled with what to do with this for a few minutes before I realized that if you colored in all the letters associated with ‘Red’ in the 5 by 5 grid, you got a letter. Repeating this process for each color gave the final answer.
Looking at our sheet, we realized there was only one puzzle left of the page. We weren’t sure whether it was the meta or another puzzle, but we decided to take a break (since we had plenty of time) and look at all the puzzle’s we had to see if we could figure out anything about the meta. We knew that each puzzle had dates on them which we hadn’t used yet. Wesley had gotten bored during Meteor Impact and had started looking at them and had realized that the months were distinct. We had started writing down all the puzzles, answers, and dates on one sheet for easy reference later when Grant noticed that the puzzle completion stamps we’d been getting had varying numbers of segments. Using this as an index gave us something that looked like it could be a real place, once we realized there might be an abbreviation involved. So we wondered if we might have a pure meta instead of the traditional shell meta after all. In any case we decided to head off and look at the last puzzle to see if it gave the letter we thought it would.
Gamma Ray Burst: In retrospect, we probably could’ve backsolved this since we thought we knew where the answer would lead us. It was a short puzzle based on alpha and beta decays on elements. We lost a lot of time because we couldn’t find one of the elements in the grid. Not really much to say about this puzzle, it was pretty straightforward.
Paradox (Metapuzzle): So they had an interesting gimmick for the meta. You could wager points before taking the puzzle. If you did so, they would not confirm steps or the answer. I was a little hesitant, but we decided to wager full points, which was the right decision. I regret not just turning in our answer without looking at the puzzle. It quickly became apparent, that the puzzle was just coded instructions telling us to do what we already did. The gimmick, as I noticed from the title, was that each set of 3 pictures could have a word appended to it to make a common phrase (e.g., Bubble, Merge, and Bucket SORT). We sat around for about an hour playing cards and cheering on a nearby team that was trying to figure out the meta. Alas, they never did.
Best Puzzle: Pandemic, despite the fact that we sucked at it.