|
It may have been executed in an ethical way, but there was nobody responsible for making sure it was executed ethically, it was just folks loving around with settings. Unethical behavior here would include things like: - Not collecting the right data, so you're just giving people randomly unequal service for no benefit - Keeping the experiment going after it is obvious that some settings are better and LinkedIn is keeping some people in a worse state - Making sure this experiment had to be executed the way it was done, and not say, letting people set their own levels, or making more educated answers based on existing data and publications There's other potential issues but that's what I see
|
#
¿
Sep 26, 2022 00:10
|
|
|
|
| # ¿ May 20, 2024 23:44 |
|
|
Owling Howl posted:Musk is a spiteful prick with a temper. Firing them is expensive and dumb but having them marched out under escort is plainly just to serve his own vindictiveness. Elon walking through a backyard full of rakes labelled "Obviously dumb legal fights". I don't think he had to pay any punishment over the Buy Twitter lawsuit, but escorting folks out like criminals seems like the kind of thing that will cost him. Its also a good point that Twitter doesn't really have code problems. They aren't losing hundreds of millions because their system is unstable or tough to improve. They have business problems.
|
|
#
¿
Oct 30, 2022 14:25
|
|
|
Lead out in cuffs posted:Yep. And it sounds like it's basically that all over: I hope this plan does well, because as a manager I would love for "Don't wait for an opportunity to be handed to you" to be the new style.
|
|
#
¿
Nov 3, 2022 13:01
|
|
|
Giant Metal Robot posted:The other amazing part of this is that the banks that loaned him money wanted to resell shares in those loans those to other people. But Elon probably can't make the $1 billion in interest that he owes on the loans, so why would anyone want to buy that debt. I thought he used Tesla stock for collateral? Another company that lost money hand over fist for a very long time.
|
|
#
¿
Nov 3, 2022 16:11
|
|
|
Lead out in cuffs posted:Yeah I'm curious about the legal implications of this. Like there ~probably~ isn't a law against it, but I'd expect Tesla shareholders would be within their rights to sue Musk for pulling Tesla engineers to work on his private side-project. Odds are the engineers are taking paid or unpaid vacation time from Tesla and getting paid to consult at Twitter. There are illegal ways to do this as well, but this under $1mil in a $44bil deal, the cost of all this is lost in the noise. Any Tesla engineer Musk wants to do some dumb code review can find a job elsewhere, so he's not going to break any employment laws over this. And there is no competition issue between Tesla and Twitter.
|
|
#
¿
Nov 5, 2022 02:32
|
|
|
Volmarias posted:Is there someplace that actually has the details? I went looking but had trouble finding anything, though I'll admit I didn't look too hard. 1bil/year aka 3mil/day is what people qualified to guess about Musk's payment terms say. I think I saw that Twitter lost 250 mil last year, would have lost about the same this year but they sold something big.
|
|
#
¿
Nov 7, 2022 05:09
|
|
|
Paracaidas posted:You, an idiot a week ago: Tanking upfronts is the dumbest thing Elon could do Not afraid of the FTC, is afraid of keeping people around to ensure compliance with FTC.
|
|
#
¿
Nov 10, 2022 17:58
|
|
|
I'm curious what a lawyer has to say about Musk/Twitter's legal responsibility. They have no t really responsibility for what folks say, but it seems like that would change if they knowingly change the system so folks can easily impersonate other people. I don't know that there's ever been a comparable situation, especially given how the rules here suddenly got worse.
|
|
#
¿
Nov 12, 2022 18:13
|
|
|
Epic High Five posted:It's pretty frustrating how years later nobody is even denying that Hyperloop was just a strategy to kill public transit and get some free money to boot, but people still think this self-driving or electric car stuff is in any way fundamentally different. At least the subway-but-for-cars system never actually killed anybody. That I'm aware of at least. Musk isn't even the first freak who realized that if you make a car that is exceptionally likely to kill people in a way you can reliably get away with it will sell like crazy to Americans. What the gently caress? Did Hyperloop actually get any public money? How did it have any strategy beyond milking VC?
|
|
#
¿
Nov 14, 2022 04:03
|
|
|
Leon Trotsky 2012 posted:Very cool link! Thanks for sharing. Eli Lilly? Speaking of which, how many engineers will be left at Twitter after the loyalty oath is due? I wouldn't be surprised if only10% signed, but realistically I don't think he'd offer 3 month severance to most of the company.
|
|
#
¿
Nov 17, 2022 05:05
|
|
|
Absurd Alhazred posted:I would take the severance, but I also would expect to get most of it eaten up by the preceding creditors at the bankruptcy. Fair, but that's also the better option compared to working at a company going through bankruptcy.
|
|
#
¿
Nov 17, 2022 05:16
|
|
|
How has Twitter been holding up through the World's Cup without engineers?
|
|
#
¿
Nov 22, 2022 02:10
|
|
|
I've lived in Boston for a long time and I've never had snow tires. Cars have been Accord and Civic, never any trouble outside a bit of skidding when it was actively blizzard out there. I think a big part is that South East New England (including greater Boston) does a great job ploughing. The roads are absolute poo poo condition but the ploughs are out an hour before the snow starts.
|
|
#
¿
Dec 7, 2022 23:00
|
|
|
You can post where you are on twitter, but everyone who retweets will be banned
|
|
#
¿
Dec 16, 2022 17:11
|
|
|
Silly Burrito posted:You can get it certified?????? Let me tell you about horse semen! Horses are bred, and some folks think that breeding is important. How many generations of race winners they have is a big one but there are other traits and you also have to make sure that the horses aren't closely related as well. The super star breeding doesn't happen because horses just get crazy at the bar after a big race, it turns out that the folks who own horses have a lot of disposable income, so there's a whole business attached to the business end of the horse. First they extract the semen by essentially having the horse screw an inflatable horse, with a whole team of assistants around to make sure it happens safely. Then they freeze the semen, ship it around the world, defrost it, make sure it is still viable, and artificially inseminate the mare. There is a whole business around each step in this process, including a business to certify that the semen came from a horse with the characteristics you, the horse semen buyer, are looking for, and it was alive and treated correctly when they shipped it. There's probably a similar system for other types of livestock (and maybe zoo animals?), but I imagine that horses are where the big money is.
|
|
#
¿
Dec 18, 2022 00:07
|
|
|
PhazonLink posted:arent the top animals in various rich people animal hobbies so inbred that there's a tiny movement to maybe stop being so inbred? I don't know a ton about that problem, but the horse semen business is big money, so most folks, especially the ones who want to breed race winners, would try to avoid inbreeding. I can imagine that some idiot millionaire has said "Breed them with Horse X, I don't care if they're cousins", and it's probably not great that this is a legal thing to do. I would think the problem is bigger on family farms, where they've had a closed system of 20 horses hanging out for generations. Also pets like the Scottish Fold cats, where its a small breed and a lot of the good pet owners spay them, so the breeding pool must be pretty small.
|
|
#
¿
Dec 18, 2022 00:59
|
|
|
If his specialty and interest is in squeezing processing power out of GPUs, it seems like he'd try to produce more efficient platforms for running AI/NN/ML systems, not necessarily getting into the training and algorithm part.
|
|
#
¿
Dec 18, 2022 16:41
|
|
|
Mega Comrade posted:It is the Microsoft way. Still convinced they will gently caress up GitHub at some point. GitHub has been doing really well under Microsoft. Their GitHub actions seem to have gotten much more useful since the purchase. We used to use Travis, which essentially specialized in just actions, and they've fallen apart
|
|
#
¿
Dec 22, 2022 19:11
|
|
|
What's the connection? This is the first time I've heard Spotify and Genius in any connection with each other. Is there a reason to hate Spotify or is it just 'haha red numbers'.
|
|
#
¿
Jan 7, 2023 00:50
|
|
|
Ruffian Price posted:what was your thought process reading about Bird leaving scooters everywhere Thanks for asking!
|
|
#
¿
Jan 7, 2023 15:46
|
|
|
There probably is a space for a super-chatbot to help with help line stuff, but it will take a lot of the right work to get there. Crisis hotline stuff especially is emotionally draining work, and a lot of it is just listening to folks, making comforting statements, and following a loose script, but the time you do need to speak up are hugely important and you can't afford to gently caress that up. There might be a useful place for the chatbot to help in training the helpline folks, or maybe provide helpful feedback after the fact, but there would absolutely need to be privacy and it tech should just forget about trying to make money off of this.
|
|
#
¿
Jan 7, 2023 21:11
|
|
|
Vegetable posted:This seems overblown. It’s akin to providing two playbooks of template responses to two sets of volunteers and you monitoring their work to see which playbook works better. In a near-ideal world it definitely could help, but if folks move fast and break things then folks can lose trust in the whole institution of therapy, and that is something we should prevent, and something the existing guard rails are there to prevent. IMO, folks should always know if they're talking to a person or a robot. It's just polite. They should have told folks they may be talking to a robot, and let people refuse and talk to a real person. Privacy issues are also real, folks should be able to opt out.
|
|
#
¿
Jan 8, 2023 04:01
|
|
|
Chatbots could be great at translating law into common language and making it more understandable for folks, they're especially good at taking 2 common areas and combining them into a single answer. But, answers would need to be reviewed, and the whole thing would need to be retrained and tested any time the law changes.
|
|
#
¿
Jan 22, 2023 16:24
|
|
|
So, IANAL, but if I stand in front of a court and say junk, I'm the one responsible for that, even if a robot is whispering in my ear telling me to say stuff. If my buddy Tony the not-lawyer says "just fart loudly and you're free to go", the judge shouldn't care, I'm the guy responsible for doing what I do in the court. If Tony is my lawyer and he says that, and he is representing me, then there's a chance he does take a lot of the blame. If a chatbot is whispering telling me what to say, is it "representing" me? Not in any legal way I think. And them saying they will tell you what to say but not actually represent you or be responsible, that sounds like it will break some very old laws about misleading advertising, and it is misleading in a way that could cause repercussions including physically throwing people into jail. And then there are legal protections that you get when talking to your lawyer, but you wouldn't when talking to your chatbot, so anything you say to it could be subpoenaed.
|
|
#
¿
Jan 26, 2023 18:05
|
|
|
I think it breaks down to: the chatbot should be treated as reference material, not an attorney.
|
|
#
¿
Jan 26, 2023 18:28
|
|
|
ChatGPT is like a guy with a C+ in literally everything, but with absolutely unshakeable confidence, and that confidence ends up making the tool very risky to use. I've used it for technical questions and it's great for getting simple answers that pull in common issues from a variety of places, Wrong answers on a new tool would be fine if it could say "I'm 45% sure the capital of Turkey is Istanbul, 30% sure it is Ankara, 25% it's something else". Something like that could let it replace StackOverflow, because it would give a clear indication of when you need to dig deeper for the right answer. E: I'm talking here about using it for answering questions, which is a different use case from using it to deliberately generate bullshit, although the use cases inherently point to the same tool StumblyWumbly fucked around with this message at 17:30 on Jan 29, 2023 |
|
#
¿
Jan 29, 2023 17:27
|
|
|
Main Paineframe posted:So far, it seems like the only professionals and subject matter experts that can benefit from using ChatGPT are professional spammers, marketing experts and propagandists, and career scammers. No, it has capabilities beyond writing stories. It is useful at generating sequences or code for common but niche applications like weird CI/CD actions, and I've heard its useful at translating between conversational language and weird AWS permission structures. I've actually gotten better results from ChatGPT than I did out of a particularly difficult intern. Important thing here is that the results need to be tested. It's often pretty close but broken, but it can still save time depending on the task, the problems are not subtle.
|
|
#
¿
Jan 29, 2023 19:34
|
|
|
Captain_Maclaine posted:Also, lol at your medicine example. Inoculation predates industrialization in some parts of the world by centuries (China and India as only two examples), and was in those places supported by government funding and mandate. Super weird that the population has increased dramatically post industrialization. Must be a giant coincidence. This is the perfect place for these discussions because the only counter to ChatGPT generated bullshit is the same counter to other generated bullshit, figure out what the argument is saying, weigh it against facts, history, and goals, and see if it makes sense. The argument here, that inoculation existed pre pharma, doesn't really address the point that medicine and medical treatment has advanced hugely since industrialization.
|
|
#
¿
Jan 29, 2023 20:41
|
|
|
Main Paineframe posted:It's not that Luddites didn't like their jobs being replaced with machines. It's that they didn't like being forced into unemployment or into lovely factory jobs that paid far less and had even worse working conditions. This is a much more reasonable and actionable line of argument than folks saying technology should be stopped because it always causes war and suffering.
|
|
#
¿
Jan 29, 2023 20:55
|
|
|
pumpinglemma posted:I think literally no-one itt is saying that. Someone brought in a comparison to the luddites and then people started pointing out that for the duration of their own lifetimes the luddites were right. (To be clear, I agree this was due in large part to poo poo like the enclosure acts rather than the technology itself, but our current governments are hilariously evil and corrupt as well so that's not a great counter-argument from the viewpoint of AI.) Epic High Five posted:An assault rifle is just a tool, too. Same with bombs and rat poison. This is where the whole derail started from, I think we've all learned a lot since then. shoeberto posted:How many folks itt debating ChatGPT have been involved in shipping a product to market based on ML/AI/big data? I've been involved in some big data stuff and looking at ML solutions. It's tough!
|
|
#
¿
Jan 29, 2023 21:30
|
|
|
It can make some very weird jumps. A few years ago, I started getting ads for paramilitary gear after I searched for Hawai'ian shirts. That was depressing.
|
|
#
¿
Feb 6, 2023 05:28
|
|
|
I love the idea that conspiracy nuts might think Jira is a digital equivalent of the smoke filled back room full of intricate illegal plans, when really its where overworked middle management goes to ask "Is this still a problem? Can I close this issue?"
|
|
#
¿
Feb 9, 2023 04:03
|
|
|
Karia posted:My dad works at Google and is getting ready to retire. Honestly we were all really excited to see if he'd get laid off, they were giving pretty good severance packages and it'd be a great way to kick off retirement. Naturally, he wasn't on the list, even though a bunch of his teammates were. My sneaking suspicion is that they figured they could save themselves the severance since he'll probably be quitting within the next six months anyway.
|
|
#
¿
Feb 10, 2023 15:45
|
|
|
Is Google's money still largely from advertising, YouTube (advertising), and I guess Android? Gmail and Google flights, Google docs are all handy but it seems like a ton of effort is just paid school projects
|
|
#
¿
Feb 11, 2023 00:32
|
|
|
The thing with HR is that they get brought in spend most of their time dealing with highly lovely situations or insurance (which is poo poo, that's the joke). Even when they're involved in hiring, they deal with a lot of lovely candidates before they talk to the one who will actually get hired. Definitely don't go to HR if you want help finding a compassionate resolution to a problem, but also I see how they got there.
|
|
#
¿
Feb 11, 2023 04:02
|
|
|
I have trouble believing that really happened
|
|
#
¿
Feb 21, 2023 14:23
|
|
|
A tenet of crypto is that they limit the amount of coin by requiring a certain amount of processor time/energy to go into making their useless junk. They'd love to use less power but there just aren't many options, and there's arguably a weird mover disadvantage to making more efficient mining gear. This isn't the thread to talk about it, but ChatGPT does have the potential to do something good. Once trained, it doesn't use much energy, and I'm sure folks will work on efficiency because they want to run it at a profit. Plus, when it works it would save energy with less clicking through the wrong websites or fumbling through junk. Using it to do useless chatbot stuff will always be a waste tho.
|
|
#
¿
Feb 21, 2023 21:17
|
|
|
Epic High Five posted:I think it can be useful, I just don't think it will be, and I think this strongly enough that I don't feel compelled to pretend it's going to be doing anything but clogging stuff up and putting people out of work and into precarity. If its useless and putting people out of work, were those people also useless or how does that work?
|
|
#
¿
Feb 21, 2023 22:47
|
|
|
I get the feeling folks who like small phones end up replacing their phone less than folks who like big ones. I guess I assume if you have a smaller phone you use it for less, its less likely to break, and you're not caught up in getting the latest. If that's the case, even if the population of small phoners is like 40% of the people, the annual unit sales could be like 20% of the total market
|
|
#
¿
Feb 23, 2023 21:23
|
|
|
|
| # ¿ May 20, 2024 23:44 |
|
|
Ghost Leviathan posted:I feel like it's missing the part where Japan was basically ordered by the US to commit economic suicide before they became a threat to American hegemony.
|
|
#
¿
Mar 2, 2023 03:36
|
|