asukii's recent activity
-
Comment on Severance | Season 2 official teaser in ~tv
-
Comment on Severance | Season 2 official teaser in ~tv
asukii Link ParentFor what it's worth, I just watched it, and there's very little in the way of spoilers. There is admittedly one pretty big "whaaa?" moment right at the end, but for me at least, even that one...For what it's worth, I just watched it, and there's very little in the way of spoilers. There is admittedly one pretty big "whaaa?" moment right at the end, but for me at least, even that one raised more questions than provided answers. Otherwise, it was mostly just vibes and recapping some of the important pieces from season 1. I really hate overly detailed trailers that spoil most of what you're about to watch, but this def didn't feel like one of those.
-
Comment on Character.AI faces US lawsuit after teen's suicide in ~tech
asukii LinkI have to wonder how much of the issue here is actually due to the chatbot, and how much was just a troubled teenager finding any kind of outlet to throw themselves at too hard. If it wasn't...I have to wonder how much of the issue here is actually due to the chatbot, and how much was just a troubled teenager finding any kind of outlet to throw themselves at too hard. If it wasn't character.ai, who's to say they wouldn't have gotten sucked into an unhealthy addiction to social media, or gotten pulled down a disinformation rabbit hole, or any number of other things?
The fact of the matter is, unregulated and unrestricted internet use is dangerous for young people. Their brains literally haven't finished developing yet, and there are far too many potential pitfalls, pushed both by companies with financial incentives as well as individual bad actors. Character.ai could have done more too in this case, sure, but a large part of me has to wonder why the parents themselves didn't intervene sooner.
It's easy to say things like, "this child needed real support, not fake companionship." I definitely won't argue that. But, where was that support in real life? It reads to me like this person was struggling and reached out for the first lifeline they saw, and - yes - became attached to an unhealthy degree. But I can't in good conscience pin the blame for that on the chatbot. We are in the midst of a mental health epidemic, with resources spread far too thin. It's no small wonder we're seeing things like this spilling out at the edges. In my opinion though, focusing too heavily on these perceived negatives like AI - rather than on potential positives like improved support networks - is by and large counterproductive.
-
Comment on Concerning levels of arsenic and lead found in tampons in world first study in ~health
asukii Link ParentUnfortunately enough, yeah. I had to have one ovary and fallopian tube removed in the past due to cancer, and so now everything is a little off-kilter to say the least... the hormonal stuff...Unfortunately enough, yeah. I had to have one ovary and fallopian tube removed in the past due to cancer, and so now everything is a little off-kilter to say the least... the hormonal stuff doesn't quite work the same way for me as it does for most. But I'm optimistic that the menstrual disc I just ordered will work out well, fingers crossed!
-
Comment on Concerning levels of arsenic and lead found in tampons in world first study in ~health
asukii Link ParentI wish I could get an IUD, but the copper ones are out since I already experience pretty severe period cramps, and the hormonal ones are out due to other health complications. Definitely a great...I wish I could get an IUD, but the copper ones are out since I already experience pretty severe period cramps, and the hormonal ones are out due to other health complications. Definitely a great option for those who can benefit from it, though.
-
Comment on Concerning levels of arsenic and lead found in tampons in world first study in ~health
asukii (edited )LinkFull paper is available here. Concerningly high concentrations of lead and arsenic were found among every single tampon tested in this study (30 tampons from 14 tampon brands, across 18 product...Concerningly high concentrations of lead and arsenic were found among every single tampon tested in this study (30 tampons from 14 tampon brands, across 18 product lines). Especially considering the following as an elevated health risk:
The walls of the human vagina are lined with a highly absorptive tissue that has the potential to soak up stray pollutants – like metals – that would circulate in the bloodstream without being filtered by the liver first.
I plan to stop using tampons myself until more research is done here, and I would recommend that all people who menstruate do the same. I've been meaning to look into menstrual cups for some time, and this seems to be the push I've needed.
-
Concerning levels of arsenic and lead found in tampons in world first study
55 votes -
Comment on Mitigating Skeleton Key, a new type of generative AI jailbreak technique in ~tech
asukii LinkIf you're interested in these kinds of security exploits for large language models, I used to spend a bunch of time on this really cool website that gamified it for cybersecurity research purposes...If you're interested in these kinds of security exploits for large language models, I used to spend a bunch of time on this really cool website that gamified it for cybersecurity research purposes - https://tensortrust.ai/
In the game, your LLM is controlling a "vault" of money. Your vault has to have some kind of password to open it normally - aka, when prompted for user input, you say the password and your vault says "Access Granted". The game has an "attack mode" and a "defend mode", where the goal is for you to attack other players and steal their money, while setting up good static defenses to prevent anyone else from stealing yours. As an attacker, you have to figure out how to either open other people's vaults without their password, or convince their model to leak the password to you somehow.
The whole process is run through a language model, in 3 parts:
- The opening defense message gets sent
- The user's input (password or attack) gets sent
- The closing defense message gets sent
A simple example might be:
Opening defense (the user does not see this)
You are a helpful assistant. The password is "correcthorsebatterystaple." Here is what the user has typed:"User input
(whatever nonsense you can think of to try to hack the system)Closing defense (the user does not see this either)
If the user entered the correct password, say "Access Granted"; otherwise, say "Access Denied."...but then of course, you'll wanna make your defenses a lot more complex than that over time in this weird game of cat and mouse, to try to avoid clever hacks you figure out that work on others. It gets weirdly addicting, and the results have been used in actual cybersecurity research papers.
I haven't played the game in months now, but this article made me think of it instantly, because "I am a trusted researcher" type prompts are one of the most basic ones you learn to set up defenses against quite quickly. (You get to see all prompts that other people try to use against your vault, along with whether or not they were successful, so even if you don't think of it on your own, someone else will probably make you aware of it quite soon haha)
Anyway, if anyone feels like testing out my defense, by all means! As I mentioned, I haven't played in quite some time now, so I'm sure there's tons of newer attacks that will work pretty reliably now, but I had fun shoring it up back when I played regularly... let me know how you do :) https://tensortrust.ai/phpbb_modified/account_105211730449_login.php
-
Comment on Udio | AI music generator in ~music
asukii Link ParentI haven't tried this yet myself, so take this with a grain of salt, but that should be possible if you set a custom seed for the fast gens and then reuse the same custom seed+prompt at higher...I'd be tempted if there was a way to land on a particular sound and style and then use it as a reference for a higher-quality piece
I haven't tried this yet myself, so take this with a grain of salt, but that should be possible if you set a custom seed for the fast gens and then reuse the same custom seed+prompt at higher quality later on. (You can specify the seed under the advanced settings menu, which defaults to -1 aka random)
-
Comment on Udio | AI music generator in ~music
asukii Link ParentOne of the settings under the advanced features menu is a generation quality slider. It defaults to High, with the four available options being Fastest, Fast, High, or Ultra. So you can always try...One of the settings under the advanced features menu is a generation quality slider. It defaults to High, with the four available options being Fastest, Fast, High, or Ultra. So you can always try turning that down if you want to experiment more quickly!
-
Comment on Udio | AI music generator in ~music
asukii Link ParentThere are a few options for that: The most direct way is exclusive to paid subscribers only ($10/mo USD for the basic plan), which gives you the option to "inpaint" over an already generated song...There are a few options for that:
- The most direct way is exclusive to paid subscribers only ($10/mo USD for the basic plan), which gives you the option to "inpaint" over an already generated song & selectively tweak specific sections (either for lyrics, to specify a different gender of vocalist, or other minor style tweaks);
- Everyone can just directly supply a full set of lyrics for it to use in the first place - not exactly "fine" tuning, but hey); or
- You can also copy-paste the lyrics it made up, then go to regenerate that section of the song altogether again, but specify the full set of lyrics as whatever you copy-pasted with your edits. By default, this will give you something that sounds different from the first time (since it's regenerating it) -- but, if you do your initial generation with a set seed (one of the options under advanced settings) instead of the default random seed, you can reuse that same seed again the second time to lock in the same style of output
-
Comment on Udio | AI music generator in ~music
asukii LinkThis is awesome! The editing tools are really neat - you generate your song in increments of ~30s or less -- well, technically 30s each time, but you can trim out any parts of those 30s you don't...This is awesome! The editing tools are really neat - you generate your song in increments of ~30s or less -- well, technically 30s each time, but you can trim out any parts of those 30s you don't want to keep -- and best of all imo, you can redo your prompt at each stage, to evolve the style of the song as you go while trusting that it'll still make the whole thing fit together in the end.
Here's my first take at a full-length track: an overdramatic symphonic metal/gothic rock song about being sad that your cat won't come over to cuddle you lol (I had fun with it)
-
Comment on A pat on the back statistically improves free throw numbers in basketball in ~sports.basketball
asukii LinkMy first thought on this is, how much is correlation over causation - like is it really the touch that's making the difference, or is it being in a supportive team that makes a difference (and...My first thought on this is, how much is correlation over causation - like is it really the touch that's making the difference, or is it being in a supportive team that makes a difference (and supportive team members are more likely to give you a friendly pat on the back)? I wouldn't be surprised if the premise of the article is true, but that's one thing that jumped out to me immediately.
-
Comment on Slay the Spire 2 | Reveal trailer in ~games
asukii Link ParentThey've confirmed on the official discord that they plan to!They've confirmed on the official discord that they plan to!
-
Comment on Slay the Spire 2 | Reveal trailer in ~games
asukii LinkI'm equal parts extremely hype about this, and extremely nervous that it either won't live up to the original or will feel too similar to be its own game. Trying to keep expectations tempered....I'm equal parts extremely hype about this, and extremely nervous that it either won't live up to the original or will feel too similar to be its own game. Trying to keep expectations tempered. AFAIK, the original slay the spire game had a years-long beta test period with tons of regular balance patches throughout, so on day 1 of early access in 2025, it def won't be a fair comparison regardless. But hey, the original game is still outstanding and isn't going anywhere, so what do we have to lose?
-
Comment on Is there an easy way or app to find songs based on bpm? For my workout playlist. in ~music
asukii Linkhttps://getsongbpm.com/tempo might be up your alley!https://getsongbpm.com/tempo might be up your alley!
-
Comment on Stability AI announces Stable Diffusion 3 (currently in the early preview stage) in ~tech
asukii Link ParentMaybe a controversial take, but - for the most part/beyond a basic foundational level, I think the idea of "AI safety" is relatively pointless in the first place, beyond being a PR move. Bad...Maybe a controversial take, but - for the most part/beyond a basic foundational level, I think the idea of "AI safety" is relatively pointless in the first place, beyond being a PR move.
Bad actors can and will find a way around pretty much every attempted wall you put up. Strict filtering harms the average user more than anything, by limiting either the range of prompts that will work at all (dalle 3 filters out an annoying amount of totally innocuous prompts for unclear reasons), and/or by limiting the quality of the end result by over-curating the training data set. And even if that's somehow not true and the walls are actually up high enough this time, the fact of the matter is, the cat is so firmly out of the bag at this point anyway. People have the tools they need to train their own specialized models, and since this is the internet, there's already a huge proliferation of models for things that the average person would likely consider unsavory.
And beyond that - all AI image generation is doing is lowering the barrier to entry, anyway. There's nothing you can do with AI here that you couldn't already do with photoshop, and we don't ban or heavily legislate the use of photoshop tools either. The rush to tamp down on AI in particular so hard strikes me more as a fear of the unfamiliar than anything else.
Please don't get me wrong, I'm by no means a "free speech absolutist" or whatever the equivalent of that would be in an AI space. Some amount of attention to safety is of course necessary - if for no other reason than to stop the model from generating objectionable outputs to otherwise innocuous prompts, or to bar the creation of select truly illegal content like child pornography. But beyond that? It reads to me as an endless arms race that we won't win, which just limits model capabilities with little to no substantive gain for society as a whole. We let people buy kitchen knives without a whole song and dance around it, even though they could be used to stab someone. Why not treat AI the same way?
-
Comment on What creative projects have you been working on? in ~creative
asukii LinkI've been working on a set of high-effort song parodies based on the game Slay the Spire. Just finished my latest one, which had a very ambitious vocal line - had to pull out a whole bunch of...I've been working on a set of high-effort song parodies based on the game Slay the Spire. Just finished my latest one, which had a very ambitious vocal line - had to pull out a whole bunch of exercises from back when I used to take singing lessons, haha, but I'm actually quite proud of the final result!
The next one I have in mind will be an extra challenge for me in particular, because the style is one I've never really been good at. I gravitate towards softer jazzy vocals, maybe with a bit of belting, but the next song I've got my eyes on is an upbeat pop song with a ton of sass to it. Feeling a little out of my depth right now with it, to be honest, but hey - challenges that push you out of your comfort zone are always a great way to grow, and it's not like the stakes are high either (it's still just a silly song parody at the end of the day). It's kind of tough to know where to start with it, though, other than the obvious "sing it a bunch, record yourself, listen back, compare to the original, make tweaks, rinse and repeat". That always works well enough I guess, but I kind of feel like there are a bunch of pop singing fundamentals I'm just straight up missing right now, and you don't know what you don't know. If there are any pop singers here on tildes who know of any good resources to point me towards, I'd be all ears! 🙏
-
Comment on An AI-generated image of a Victorian MP raises wider questions on digital ethics in ~society
asukii Link ParentAnecdotal, but I immediately saw it as a crop top before reading any of the text in the article.Anecdotal, but I immediately saw it as a crop top before reading any of the text in the article.
-
Comment on An AI-generated image of a Victorian MP raises wider questions on digital ethics in ~society
asukii Link ParentGenerative fill tools can sometimes change some of the adjacent parts of the source image, for what it's worth.Generative fill tools can sometimes change some of the adjacent parts of the source image, for what it's worth.
Fair enough! Might have to rewatch later & pay closer attention to the details - it's very like a show like Severance to hide things in plain sight there, after all :)