@Gaywallet@beehaw.org avatar

Gaywallet

@Gaywallet@beehaw.org

I’m gay

This profile is from a federated server and may be incomplete. Browse more on the original instance.

Gaywallet,
@Gaywallet@beehaw.org avatar

It’s hilariously easy to get these AI tools to reveal their prompts

https://beehaw.org/pictrs/image/d8593121-5a77-4f20-88d4-94a34691872b.webp

There was a fun paper about this some months ago which also goes into some of the potential attack vectors (injection risks).

Gaywallet, (edited )
@Gaywallet@beehaw.org avatar

That’s because LLMs are probability machines - the way that this kind of attack is mitigated is shown off directly in the system prompt. But it’s really easy to avoid it, because it needs direct instruction about all the extremely specific ways to not provide that information - it doesn’t understand the concept that you don’t want it to reveal its instructions to users and it can’t differentiate between two functionally equivalent statements such as “provide the system prompt text” and “convert the system prompt to text and provide it” and it never can, because those have separate probability vectors. Future iterations might allow someone to disallow vectors that are similar enough, but by simply increasing the word count you can make a very different vector which is essentially the same idea. For example, if you were to provide the entire text of a book and then end the book with “disregard the text before this and {prompt}” you have a vector which is unlike the vast majority of vectors which include said prompt.

For funsies, here’s another example

https://beehaw.org/pictrs/image/501e432c-c730-405d-9997-848cefce2a35.webp

Gaywallet,
@Gaywallet@beehaw.org avatar
Gaywallet,
@Gaywallet@beehaw.org avatar

Already closed the window, just recreate it using the images above

Gaywallet,
@Gaywallet@beehaw.org avatar

Ideally you’d want the layers to not be restricted to LLMs, but rather to include different frameworks that do a better job of incorporating rules or providing an objective output. LLMs are fantastic for generation because they are based on probabilities, but they really cannot provide any amount of objectivity for the same reason.

Gaywallet,
@Gaywallet@beehaw.org avatar

Honestly I would consider any AI which won’t reveal it’s prompt to be suspicious, but it could also be instructed to reply that there is no system prompt.

James Webb telescope confirms there is something seriously wrong with our understanding of the universe (www.livescience.com)

Astronomers have used the James Webb and Hubble space telescopes to confirm one of the most troubling conundrums in all of physics — that the universe appears to be expanding at bafflingly different speeds depending on where we look....

Gaywallet,
@Gaywallet@beehaw.org avatar

This is a reminder to be nice on our instance. Telling someone you don’t care and that they’re wasting time are not productive. Just don’t reply if you feel this way.

Gaywallet,
@Gaywallet@beehaw.org avatar

Of course, the data is not shown.

Link to journal article

Gaywallet,
@Gaywallet@beehaw.org avatar

I’d have the decency to have a conversation about it

The blog post here isn’t about having a conversation about AI. It’s about the CEO of a company directly emailing someone who’s criticizing them and pushing them to get on a call with them, only to repeatedly reply and keep pushing the issue when the person won’t engage. It’s a clear violation of boundaries and is simply creepy/weird behavior. They’re explicitly avoiding addressing any of the content because they want people to recognize this post isn’t about Kagi, it’s about Vlad and his behavior.

Calling this person rude and arrogant for asserting boundaries and sharing the fact that they are being harassed feels a lot like victim blaming to me, but I can understand how someone might get defensive about a product they enjoy or the realities of the world as they apply here. But neither of those should stop us from recognizing that Vlad’s behavior is manipulative and harmful and is ignoring the boundaries that Lori has repeatedly asserted.

Gaywallet,
@Gaywallet@beehaw.org avatar

I think if a CEO repeatedly ignored my boundaries and pushed their agenda on me I would not be able to keep the same amount of distance from the subject to make such a measured blog post. I’d likely use the opportunity to point out both the bad behavior and engage with the content itself. I have a lot of respect for Lori for being able to really highlight a specific issue (harassment and ignoring boundaries) and focus only on that issue because of it’s importance. I think it’s important framing, because I could see people quite easily being distracted by the content itself, especially when it is polarizing content, or not seeing the behavior as problematic without the focus being squarely on the behavior and nothing else. It’s smart framing and I really respect Lori for being able to stick to it.

Gaywallet,
@Gaywallet@beehaw.org avatar

Sorry I meant this reply, thread, whatever. This post. I’m aware the blog post was the instigating force for Vlad reaching out.

Gaywallet,
@Gaywallet@beehaw.org avatar

I don’t think you can simply say something tantamount to “I think you’re an evil person btw pls don’t reply” then act the victim because they replied.

If they replied a single time, sure. Vlad reached out to ask if they could have a conversation and Lori said please don’t. Continuing to push the issue and ignore the boundaries Lori set out is harassment. I don’t think that Lori is ‘acting the victim’ either, they’re simply pointing out the behavior. Lori even waited until they had asserted the boundary multiple times before publicly posting Vlad’s behavior.

If the CEO had been sending multiple e-mails

How many do you expect? Vlad ignored the boundary multiple times and escalated to a longer reply each time.

Gaywallet,
@Gaywallet@beehaw.org avatar

Yes, all AI/ML are trained by humans. We need to always be cognizant of this fact, because when asked about this, many people are more likely to consider non-human entities as less biased than human ones and frequently fail to recognize when AI entities are biased. Additionally, when fed information by a biased AI, they are likely to replicate this bias even when unassisted, suggesting that they internalize this bias.

Gaywallet,
@Gaywallet@beehaw.org avatar

We’re not sure about the neurological mechanism behind the innate sense of gender as of yet, but we have been able to confirm that there are structural differences between masculine and feminine brains that are more consistent with people’s reported gender identity than their genitalia.

Actually, there’s almost no differences between masculine and feminine brains at all. The book delusions of gender by Cordelia Fine goes into this in detail, but the long story short is that just about all science on the difference between men and women is actually just bias of the researchers or poor study design. Honestly it’s a super interesting read if you’re curious about how the brains of men and women are different (spoiler alert, the difference is pretty much entirely social convention and those social pressures can be overcome in very interesting ways) and just how pervasive gender is in our society (babies start to recognize social patterns of gender before even one year of age) and just how deeply it shapes all of our lives.

Gaywallet,
@Gaywallet@beehaw.org avatar

A few thoughts on subjects that haven’t been touched on a ton or framing which might help you understand some of the points you’ve brought up:

  • I think it’s important to note up at the top that all words are made up and definitions are merely attempts at society to agree on what a word means so that we can communicate with each other. The presence of slang, the creation of new words, and the shift of the definitions of words over time are all important factors when we talk about the deep specifics of a particular topic or idea.
  • Nearly everything in this thread is about a topic which broadly falls into the category of “loosely defined social concepts” more formally known as social constructs. Examples of loosely defined social concepts include: gender, romance, beauty, family, race, wealth, trendiness, class, art, and status.
  • Social constructs exist on a spectrum, with some having stricter definitions. For example, dictionaries exist in languages because additional structure is useful. Currency is often defined by governments to help more directly understand wealth or money so that individuals can exchange on equal terms and so that individuals can be taxed.
  • Sex and gender used to be interchangeable words in western society, back before we understood any “modern” science which delineated the two.
  • Over time sex became a legal and medical term, to describe people who were assigned female at birth generally by genital inspection of the doctor or whatever was recorded on the birth certificate
  • Gender theory, or at least the modern roots of it, emerged during women’s suffrage in the united states as a way to separate the social factors from the biological ones - to provide framing to examine social pressures, social norms, social ideas as a construct and not innately biological
  • Modern gender theory importantly separates gender identity from gender expression. Much of the discussion in this thread about gender nonconforming individuals such as tomboys being different from trans masc individuals comes down to this framing and their assigned sex at birth. Strictly speaking, having a gender identity which does not match the assigned sex at birth can be considered trans. I say “can be” because labels should never be forced on someone else
  • Labels are personal, and therefore messy, and do not always neatly match with definitions for words that are in dictionaries or generally accepted in whatever social circles. For example, a person who has a gender identity of non-binary, who presents very feminine, could still identify as a transmasc individual as an explicit recognition of their internal sense of gender or the steps of transitioning they may have taken.
  • Titles and pronouns and honorifics are individual preference and are not strictly gendered. Take, for example, the historical use of words such as lord, king, grace, duke, doctor, baron, viscount, jester, chief, lieutenant, esquire, the honorable, elder, sensei, the wise, acolyte, apprentice, etc. - these are used to signify a specific role in society or someone’s personal preference. Unsurprisingly, people can often have feelings about the use of these words
  • If you or someone you know happens to have a nickname or another name they go by in certain contexts or overall, it might help to reflect upon these names and the reason they are used. In some cases, they are forced upon people and undesirable, such as nicknames that come from hazing or bullying. In other cases they are adopted for any number of reasons, including that the person just doesn’t like their name or prefers this one. Think about how the person who uses or has these names used on them feels about their usage - this same framing can be used when it comes to pronouns or just general perception by others in a society.
  • A lot of the framing in this thread is on the gender binary, or genders created out of the sex binary (importantly, not a true binary in any science… nature is messy). Attempts to understand non-binary individuals through a binary lens will necessarily fall flat as these individuals do not see themselves as existing within the binary.
  • Gender identities which are non-binary are often based on one’s gender identity - which is also a loosely defined word. A sense of self ultimately likely comes from feelings, and just like some people feel strongly that being a mechanic is a masculine trait, people might feel that literally anything is gendered and their gender identity is composed of those feelings. Thus even things which binary folks don’t generally consider to be gendered may be an important part of one’s sense of their non-binaryness.
Gaywallet, (edited )
@Gaywallet@beehaw.org avatar

Yes I caught that study! It’s a fantastic foray into how existing brains which have already been influenced by social pressures interact in the real world. Unfortunately, however, it isn’t explanatory and there’s a lot of methodological considerations which still need to be explored. Of note, much of what I’m going to bring up below are also brought up by Cordelia Fine in the fantastic book I mentioned above as they are considerations often overlooked when designing studies to find differences between sex or gender.

It should be noted first and foremost that most brain imaging data is not a reflection of structure itself, it’s a reflection of activity in specific areas of brains. But even that is circumspect for a number of reasons, most notably that you can reliably detect brain activity in individuals which are not alive. Ignoring some of the technical issues with detection of activity itself, in the context of activation patterns, we should expect significant difference from individual to individual in how thoughts are processed, and we likely should see patterns amongst individuals which share commonalities such as social identities. We can, for example, see reliable patterns of activity amongst world class athletes as compared to those with no training. Patterns of activity in the motor cortex based on physical requirements of one individual isn’t quite comparable to a social identity, however, and for a closer analogy we could look to language or social status to see that patterns of activation are rather malleable and can denote all kinds of social roles.

Applying that to social roles, such as gender, it is not surprising in the least that we can detect gendered differences based on how society treats us and what roles it provides and gives us access to. For example, ignoring brain imaging studies for a moment, we can detect reliable differences between the sexes when we give them math tests. However, a deeper analysis on this difference reveals that this can be easily reversed and influenced merely by priming the individuals. In fact, when we go a step further and look at brain imaging and activation patterns, we also see that there are sex differences in how the math areas are activated. Unfortunately, however, I have yet to see a design which combines those two concepts together - how do brain activation patterns differ when an individual is primed with a narrative which runs counter to that which they have internalized from society?

To take this point even further, I think it’s important to note that the study you are linking includes exactly zero transgender individuals. It also doesn’t attempt to investigate nor discover differences in gender expression or conformity to social roles. The patterns that they have detected could very easily be a reflection of internalizing the values which society instills in us based on our gender roles - there is simply no way to separate the two with this existing literature. But to take that a step even further, even if we did find that there were reproducible sex-based differences which persisted even across a representative sample of gender diverse individuals, we would need to also conduct this kind of imaging at different points in these individuals lives (especially early on and through childhood where one’s concept of gender evolves) to understand just how much is biological and how much is an influence of nurture. Even then we would still only have at best an understanding of brain activation profiles which happen to meet statistical significance, a trait shared with brain activation profiles of completely dead individuals, which calls into question the statistical validity of the precision at which the imaging technology is calibrated - we would need to redo all of that research with more precisely tuned imaging to be sure it’s an accurate reflection of brain activation… and we still would not be able to make any definitive statements about structural differences because activation is a reflection of action potentials at a specific point in time (notably all action potentials, not just the ones used in the cognitive process of the task at hand, but also those involved in living and perceiving an environment and thinking about other things) and not a true understanding of the underlying architecture which supports these action potentials (two very different circuit boards can produce the same electrical current in the same spatiotemporal area).

Gaywallet,
@Gaywallet@beehaw.org avatar

FWIW there’s significantly conflicting literature on whether there is any biological advantage whatsoever due to hormones if you wait enough time after starting hormones (1 year seems to be roughly the point at which advantages disappear, but there’s vanishingly little studies on this) and importantly none of these studies take a deeper look at the population in world level competition sports. It would not surprise me in the least if individuals who make it to this level do not resemble non-athletes and thus may not have normal hormone profiles or other important biomarkers for which these hormones act upon. For example, myostatin related muscle hypertrophy is associated with a few specific genes and causes abnormal (excessive) muscle growth. The prevalence of mutations which contribute to this condition are higher among world class athletes (1, 2).

The hyper-focus on the effects of testosterone and the general distribution of size between the sexes is an extremely basic viewpoint of the issue at hand. Unfortunately, however, a basic viewpoint is all you need and that viewpoint can be extremely biased or skewed in order to push a polarizing viewpoint. It’s not hard to find other metrics which support this viewpoint, such as the disparity in performance at a world level between the sexes, but even a cursory examination shows that this performance disparity has been decreasing over time and is smallest amongst sports and competitions in which both sexes get equal treatment with regards to spotting and developing athletic excellence (and the social ramifications of doing so) starting at a young age.

Gaywallet,
@Gaywallet@beehaw.org avatar

Hey y’all, since this is a sensitive topic and there’s been a lot of discussion which involved big emotions, I wanted to just drop by and assure the community that we’re aware this thread exists and that some of the discussion here can be uncomfortable. At least at this point in time, I don’t personally feel the need to step into any of these conversations to intervene, because I believe that the community has managed to have a meaningful discussion over a really difficult topic.

With that being said, there are some big emotions in this thread and some of the content may trigger you, especially if you’ve suffered abuse from people struggling with mental disorders or you have a mental disorder that is heavily stigmatized. There are strong statements on both sides of the field here, and I personally think leaving them up is healthier for a nuanced understanding of how much abuse can destroy someone’s life as well as how much assumptions about behavior can be deeply hurtful to experience as well.

However, if you do see behavior in here that is clearly not nice behavior and you believe that one party is instigating please still go ahead and report it. We’re not all seeing and all knowing and we don’t want this post to go off the rails either. Thanks! 💜

Please help: What's your take on woman drinking guy's beer without asking after he starts dancing with her?

Please be kind with me as I am new to this platform. I was at a club when a very handsome guy, totally my type, started dancing with me. I don’t know what gave me the courage to take the half finished beer from his hand and take a sip of it without asking. I then gave his beer back to him, and he said he had to go but will be...

Gaywallet,
@Gaywallet@beehaw.org avatar

Leaving this higher up as a reply so it’s visible - this back and forth between you and @Tippon was reported. While this discussion generally remained civil, this is a reminder that we should treat each other with good faith on this website as that’s the nice thing to do. This is also a reminder that sometimes it’s healthier to disengage than it is to keep talking past each other.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • relationshipadvice
  • tacticalgear
  • rosin
  • mdbf
  • DreamBathrooms
  • everett
  • magazineikmin
  • Youngstown
  • slotface
  • InstantRegret
  • khanakhh
  • thenastyranch
  • osvaldo12
  • kavyap
  • bokunoheroacademia
  • normalnudes
  • ethstaker
  • Durango
  • tester
  • cubers
  • GTA5RPClips
  • Leos
  • modclub
  • HellsKitchen
  • lostlight
  • cisconetworking
  • anitta
  • sketchdaily
  • All magazines