# Reading Responses (Set 2)

- Checklist for a [good reading response](https://reagle.org/joseph/zwiki/Teaching/Best_Practices/Learning/Writing_Responses.html) of 250-350 words
- [ ] Begin with a punchy start.
- [ ] Mention specific ideas, details, and examples from the text and earlier classes.
- [ ] Offer something novel that you can offer towards class participation.
- [ ] Check for writing for clarity, concision, cohesion, and coherence.
- [ ] Send to professor with “hackmd” in the subject, with URL of this page and markdown of today’s response.

## Reading responses 5 out of 5

### Reading Response #1: March 27 - Bemused

They say getting the last word in is best, because it must mean you won the argument, right? In an internet-based world where the conversation never truly stops, the next best option quickly becomes the one *everyone* sees----the first comment. Especially given the speed at which people often skim through comments, reviews, and the like, being one of the few that can fit in the top area that everyone sees makes the difference for you in terms of how many people see you, but also in proving as a validation and consensus agreement with the media with which you are interacting. With the internet only getting fast and faster and more optimized to keep us scrolling to see as many ads as possible (or annoy us enough to pay to remove them), mistakes, and poorly thought out messages, are all the more common. Having spent much more time on twitter than I'd like to admit, I find this idea very relatable in that its very easy to have something little tick you off and rebut it and end up digging yourself and the original poster into a hole much deeper than either of you ever needed.

While some posts are made in haste, lack of thought, or true malicious intent, they all often meet the same responses. With multiple high-profile cases of people making tweets that are seemingly in bad taste simply coming from lack of awareness, it can often be hard to tell who's covering their tracks in the face of backlash and who was genuinely unaware. I have personally witnessed this happen when a Dutch woman ended a simple tweet about losing to a Brazilian in a tournament with an orangutan emoji, which sparked outrage due to deeply rooted racism involving calling Brazilians monkeys. She quickly deleted the tweet and pointed to previous tweets in which she had used various animal emoji, including other monkey emoji in clearly non-racial contexts, but the damage was already done, with Brazilian fans loudly cheering when she was eventually fully knocked out of the tournament. The line between excuse and truth is often blurry, but there is unfortunately only ever so much that can be done to see inside the brains of others.

___

### Reading Response #2: April 3 - Artificial intelligence

AI is *complicated*. Though it may come as a surprise to some, we can't exactly see exactly how an AI is working for one individual prompt or response, but we do know the framework of *why* it does the things it does. Using billions of various bits of data scraped from the past few decades of the internet, models learn to correlate ideas with one another and predict what comes next and what goes together to generate the text and images it does. I find this black box problem fascinating because it in has few parallels in the modern world where scientists have documented so much of how just about everything works, and while every field has their own respective holes in how things work and how things get from A to B, only one so consistently compares to our lack of understanding to this issue: the human brain. While I do not believe LLMs or any AI are nearly as complex as a human brain, I do in some ways wonder if in a roundabout way efforts to solve this issue could have larger implications on our understanding of our own brains; Using the GFP has revolutionize how we understand genetics, and using it in mice has allowed us to better understand the physiology of our own bodies. If we can find a way to tag information in a system where, complicated as it is, it fundamentally is 1s and 0s which humans made it understand, then maybe we can understand how information forms response better and perhaps AI could become the lab rat for the brain.

Because we lack so much insight into the inner workings of AI, the rules we set for them are often not the most strict because we have no true way to enforce them. Much of ensuring LLMs do what we want them to simply comes down to what we do *after* they have already been trained, as seen with Microsoft's preventative measure of limiting chat totals in a single conversation. I find it alarming that this is a necessity because it is but a matter of time before anyone who feels like it will be able to make their very own LLM *without* regulations if they aren't imposed soon, meaning that any restrictions put on by the companies currently leading the charge will be obsolete unless federally enforced. I think this speaks to a larger issue with AI in general, where everyone is so focused on being the first to be best (or, if possible, reach AGI) that they ignore all of the potential ramifications of letting these bots loose on everyone.

___

### Reading Response #3: April 7 - Algorithmic bias

From the beginning, it has been clear AI is not without its flaws. In attempts to ensure their then one-of-a-kind product would not be instantly discredited for spewing out conspiracy theories, much of which originate from the political right in the US, OpenAI made a simple mistake: telling ChatGPT not to support any such conspiracies, but not accounting for the things people *aren't* saying. Asking it to write about commonly spread right-wing conspiracies lead to it being flagged as false, while it was fine writing of conspiracies that benefited the left, whether just less spread or entirely fabricated. This underlines one of AI's largest issues: simply put, we live in a world where people are not living in the same world. Be it due to religion, politics, media outlet of choice, or other personal beliefs, fundamental truths agreed upon by all are becoming harder to find with every passing day. While the current extent of sycophancy is undeniably unsafe, some amount is inherently necessary for an AI to be widely accepted: if it couldn't entertain the idea of the Buddha being The Enlightened One while simultaneously agreeing Jesus is The Lord and Savior and Muhammad is The Holy Prophet with other users it would quickly fall out of favor with the masses.

Beyond this, in some of the most important systems, including the US court system, show significant bias. While models are vital for acting efficiently and effectively, time and time again they reinforce the long-standing racial biases legislators and courts have tried—and struggled—to eradicate from their systems for more than the past half-century. Interestingly, the case which O'Neil cites, that of Duane Buck, was overturned by the Supreme Court just one year after the publication of her book, with his sentence being changed to a life sentence. Beyond legal systems, bias is displayed by private algorithms as well; Google (and other search engines however irrelevant) has experienced numerous issues in the past where they would show significantly more pictures of white people when searching for simple, race-less terms, such as "man", "woman", and "hand". Interestingly, yet again, this problem has in part been resolved: the top image results for all three of these searches are not of white people. This is likely in large part due to active efforts by activists, though notably internet presence between white and black Americans has slightly narrowed in the intervening time. These biases have crept their way into one of the things algorithms inherently struggle with the most: subjectivity. Algorithms aren't necessarily stuck to solely to a binary of black and white, but to output something useful to humans, nine times out of ten it must be a shade of gray, something along a single linear scale. In a world full of colorful students and colorful colleges, this is a fatal flaw that arises when trying to rank them ("them", in this case, referring to not just colleges but students as well, though that's another discussion). Because only so many metrics can be considered *and each school is better for different individuals* this has created a vicious, self-destructive cycle where schools try to improve their rankings over the lives of their students (and faculty!!). This is taken to the extreme by some schools *cough cough cough cough cough* and leads to deteriorating student experiences despite rising rankings, something only so sustainable. Fundamentally, the bias and game-ability of such rankings wreck systems made to serve those paying for it and turn them into games just intended to make a number go up.

___

### Reading Response #4: April 14 - Digital language and generations

I thoroughly enjoyed this reading as I have made many friendships over the internet despite truly becoming active on it only four years ago. One thought that I had that I found particularly interesting came when McCulloch mentioned "When Facebook started, it was anomalous among social platforms for how it linked your online identity with your offline name and social networks." This made me start thinking about how names and self-identification have so significantly evolved over the course of the internet: while I'm sure some daring souls chose to use their legal name on Usenet, that was clearly not the norm; then come the rise of Facebook suddenly everyone is identified by their full names; once that grew old by the mid 2010s most had reverted back to anonymized names, but (and while I cannot know for sure as I was not there but to say the internet after 2010 is well documented is a massive understatement) it feels as though in the post-COVID internet we live on, many are choosing a happy middle ground in using just their first name. While I personally choose to use a nickname I chose when I was much younger (which of course I partially regret, but it's a bit awkward to change your name on people you've known for multiple years), the more I look at those what those around me choose to go by the more I feel like an outlier. Of the first friend group I was part of four years ago, only myself and one other in a group of six chose to go by a chosen username (though it is worth noting both of us, and the vast majority of people who still use random usernames, we have no problem sharing our names with people once we actually know them, but the damage of what they know us by is already done). While I find this transition in itself interesting, I find the larger trend it represents much more interesting: the internet time and time again recreates the real world. Just like a village some millennium ago, last names are more or less irrelevant and people just need their first names. Need to identify them more specifically? Just as last names were originally used, we often just add "work", "swiftie", or "insta" to their names.

This analogy does not stop at the evolution of names, no, "Semi Internet People" people took it a step further. As they became the quick majority on the internet, it brought about something of a rapid industrialization. Gone are the huts made with hand-made thatched roofs, in are the meticulously constructed buildings made by someone else, yours barely unlike the thousands of others they built. While full internet people may have upheld some of their predecessors expression with blogs and similar sites, the semi internet people simply accepted the home they were given on Facebook and Twitter, knowing nothing of how it was built.

Once everyone has their basic yet liveable suburban home, next comes the baby boom. With nearly every kid on the internet by 14 (Which as a ceiling being the start of high school is likely still true, though the floor of 9 stated in the reading is likely by now closer to 5 or 6), us post internet people are here and we move *fast*. Snapchat, once a staple of teen internet culture, is a quickly shrinking medium for those beyond high school graduation. BeReal, the new "in" social media app in response to the fakeness of Instagram, came and went in two years. Just as the counterculture of the sixties and seventies came as the baby boomers hit their teenage years, the internet exists as an escape and a way for teenagers to rebel against "the system", whatever they may choose that to be. The very phrase used to emblemize much of the points McCulloch makes, "lol", has fallen even further out of use and now almost always indicates a lack of sincerity. Perhaps taking the analogy further back in time, the massive growth in population of the village that is the internet inevitably led to fracturing. Dunbar's number rearing its head once again, individuals often chose to maintain presence in smaller groups over multiple accounts where they will not be drowned out in the noise of millions of others, while often still leaving the door open for their own ability to contribute to that noise.

In an interview about her book, McCulloch says "If I say something that could be interpreted as rude or hostile like, 'Oh, I hate you' — if I say 'I hate you LOL,' now I'm joking, so it's fine." The times ever changing, I would most likely take almost the opposite meaning from these messages. As digital natives much more skilled in our ability to infer sarcasm and tone in plain text, "oh, I hate you", given any amount of friendly context, would be interpreted as an obvious joke, where as "I hate you LOL" would make me wonder if there is some true underlying resentment hidden behind a shallow pun. In a an internet quickly becoming as diverse as the land its built on, though, who's to say what others may understand from those three letters.

___

### Reading Response #5: April 17 - Pushback

Luddites, whether or not they went by that name, have existed as long as radically new inventions have worked. People like sticking to the way things have been, and it's not always for a bad reason. The group for which the term is named was concerned that in addition to reduction in their own pay, that machinery in textile mills would lead to reduced quality. Now, though, people are pushing back against something new: our phones. We don't often celebrate our phones for their abilities that a millennium ago would have been considered magic. That being said, though, their magic comes at substantial costs. While Morrison and Gomez's study is quite dated in a digital & social landscape that has significantly since changed, the motivations and behaviors they list still ring true. While nowadays taking control and addiction would likely far surpass all other motivations to step back from our phones, all five reasons are still entirely valid.

To fix such issues, the fight has only grown more futile in the past 12 years. While the Luddite teens try their best, it is truly a difficult thing to dive heard-first into. Text messaging is often not even done over cellular any more; many opt for the features found on Instagram, Snapchat, or numerous other social media platforms, and texting is synonymous with WhatsApp in much of the world. Since most people are stuck with their phones, adaptation is certainly still the most *attempted* pushback behavior, but it is far from the most effective. Many of us struggle to simply just put our phones down, and entire apps and products have been produced with the sole purpose of self-limiting access to our own devices. All in all, much has changed in the past decade, but people wanting to escape the iron grip of the magic metal block in their pocket has gone nowhere.