I have been a bit confused throughout the article about what the authors mean by "leadership identity." I initially thought they meant being leaders in various spaces, but the commentary in the article almost always talks about how their leadership is inextricably tied to a person's queerness (i.e., they are only allowed to be leaders when they are advocating for themselves). Is the article discussing leadership with respect to advocating for queer rights and not other forms of leadership? Is this, in fact, a clear symptom of the problem of the marginalization of this group?

I have seen this with several marginalized groups - we are forced to overcompensate just to be heard. However, sometimes the very nature of "making so much noise" perpetuates the discrimination these groups experience. I think about friends of mine who are activists combating anti-black racism, and they can be really difficult to approach and learn from. I have also had experiences with women in male-dominated spaces who are extra tough because they feel they need to be (e.g., female border officers who are intentionally mean or aggressive as an overcompensation in response to their marginalization or the assumption that they can't be tough or mean). I am not in any way saying that I'm against activist or speaking out -- I am commenting on how unfair it is that marginalized groups are "forced to make so much noise" and that, in doing so, they may not even achieve their desired outcome.

I know people who still, today, describe queerness as a mental pathology. Prejudice runs long and deep. I think it is so important to note that people in this marginalized group arguably do struggle with more mental health challenges than those from cisgender, heterosexual groups; however, these challenges are generally a result of discriminatory treatment, not queerness itself. Mental health struggles come from minority stress, not queerness. In addition, gender dysphoria exists on the DSM5, I believe, not because being transgender is a mental illness but because the distress caused by a mismatch between gender identity and assigned sex can be clinically treated.

I also take huge issue with the comment that conflates mental illness with deviance or immorality. In my opinion, the authors of this article could have done a much better job of clarifying this observation/view. As the article currently reads, there is no pushback on the assumption that a person with a mental health disorder is deviant or immoral, just that queerness was considered a mental health disorder.

I recognize the challenge that there was only one article found during the time period (pre-1990) and I also recognize that the authors of this literature review called this a "snapshot," but I do take issue with generalizing the entire theme of a decade based on one published article. Corroboration is necessary here. Aside from mentioning that only one article met the inclusion criteria, the authors do not warn against this generalization; rather, they perpetuate it. Are they taking the absence of literature as corroboration of the "visibility" theme?

This is just a sidebar comment - I love this analogy, and I wish I had read it in time to incorporate it into my positionality paper! I think the "unraveling" is a poignant description of a critical reflexivity -- when we aim to be critically reflexive, we constantly question, shift, and reorient ourselves as we learn and interact with the world.

The reclamation of the word "queer" by the queer community is an incredible example of activism toward controlling their own narrative. Language is powerful. However, the use of this term still seems complex to me, as its meaning can change depending on who is saying it/the context in which it is used. For example, those who have intimately experienced the term as a slur may resist adopting its "new" meaning. It can also be used as a sweeping term for individuals who do not identify with the description. Overall, I think the reclamation of the term by the community creates a possibility for productive change and the dismantling of power structures, but I recognize that there is no one universally adopted interpretation (even within the community it aims to describe).

from my privileged position, I had to look up the difference in definition between repression and oppression in practice. The authors chose to use "repression" here, which is appropriate, but I would argue that so is "oppression." When the article talks about systemic mechanisms for disadvantaging (for lack of a better word) queer people, like "Don't Ask Don't Tell" or the Defense of Marriage Act, this is blatant oppression that produces repression. Excluding the voices of queer people is repression, but it arguably stemmed from blatant systemic oppression.

I think what is so powerful about this is the connection between humans, the stories and knowledge being passed down through storytelling and modeling and experience. Human connection is one antidote to AI challenges.

Who are we as humans if we are not working toward something? Where is our motivation? Our agency? Our sense of well-being and purpose?

I think it's so important that people know that there is, in fact, a choice here. We still have agency. We do not have to hand it over. And, to be clear, we do not know what the future of AI will look like.

This is true. I will say, however, that I have been pleasantly surprised when talking to some post-secondary students about their opinions of AI -- For example, they reject the idea that AI will become a better teacher than humans. This push-back is happening and is necessary.

Does this not all depend on the training data/context window? I don't think people building BigAI are thinking about solving climate change (definitely not!) or curing cancer. However, I DO think that LLMs that are responsibly trained on specialized data have a great deal of potential in solving some critical human problems.

This, I think, is the real danger. If we are using AI to generate text, not using skills of discernment/critical thinking/information literacy, we are allowing the propagation (and continued generation) of misinformation, and we may not be able to go back to correct this error.

We are already seeing censorship here, as we are not allowed to share news stories over some social media platforms. So, our opinions about news stories are not included on the platform and this leads to erasure as well.

If we allow ourselves to be fooled. We still need discernment and due diligence when using a tool.

I'm wondering if intelligent prompts can circumvent this (and AI sycophancy)? It is humans' responsibility, after all, to not allow the tool to be the boss. I don't think we have to surrender as victims, here. However, I am quite terrified that AI use by young people will result in a type of cognitive offloading that is catastrophic for critical thinking, so there's that...

If globalvoices.org is ranked 876 on the C4 list, does this not mean that the work this online community has done will make some impact on the LLMs and the data included? Or, is it heavily outweighed by the sources that rank higher?

First, I had never seen this acronym before. Second, are we only just now talking about misrepresentation on the internet? Is the worry now that people are just accepting AI output as fact? Did people not do that before with regular web searches? Is the real challenge the perpetuation of misinformation because AI models train on the data they themselves produce (which contain the misinformation and biases)?

How useful/reliable will this be as AI models write more and more of themselves? Will programmers still have control? It's interesting to think about what the potential futures are, here.

This seems obvious. What is not obvious, though, is what texts are used to train different LLMs. I'm not sure what is meant by "professional content," here, as opposed to "web content." Is only open source data available to train LLMs? If so, does this mean that scholarship is not included unless it is open source? Is this a matter of educating people that LLMs are "accurate" for things like recipes but not for things of a "professional" nature?

So, unless an LLM was trained on very specific, specialized data, it really is a glorified web search with the ability to generate information based on its context window and the patterns it detects based on statistical probability?

I'm wondering if this is truth or opinion. Since the writer does not back up their claim with evidence, I'm hesitant to accept it. Are we sure it's "Wikipedians, bloggers, and other online writers"? What proportion of online texts are from this named group? I'm not disputing that the internet is controlled by the Global North; however, I question broad claims like this. Is Zuckerman talking about the Common Crawl here? Is this the same thing as C4 mentioned later in the post? Also, if we are, in fact, getting most of our data from Wikipedians and bloggers, how is the text generated by LLMs "frequently accurate," as mentioned above?