We found that our participants were accurate in theirestimation of Internet use, and most measures werestrongly correlated. Thus, students who actually used theInternet more also reported using the Internet more.
Re: something discussed in class last week. We talked about how there's a 'social desirability' bias to some survey studies. I am surprised there wasn't more of that in this survey.
Class-related Internet use was not associated with a benefit to classroom performance.
It was also not associated with a decline in performance - in fact, according to the regression on pg 175, it wasn't even significant. Obviously the presence of laptops enables non academic internet use, but I wonder how many students made use of distractions. Point being, maybe the blame should be on non-academic internet use rather than laptops.
Participants completed a demographic survey, and on the followingday, they began to receive links to brief ESM surveys sent via textmessage using the service surveysignal.com
Sort of interesting that they did this survey via text message, when they were studying the effects of, well, being to engrossed in text messages. I wonder if that had an impact on the results?
inattention
This is sort of an aside, but I'm intrigued by this way of talking about ADHD/autism. From an inside perspective on the disorder, I don't find that it's a deficit of attention, rather that it's a deficit of attention regulation. Or in other words, I'm always focused on something, it's just not necessarily what I'm supposed to focus on, and that looks like inattention from an observer's perspective.
In Ireland the question of ownership is overwhelmingly about who owns the past, a place that remains unstable or unfinished, and one from which women have tended to disappear.
The Italian word for 'history' - historia - has an alternate definition as 'story.' I think this dual meaning can inform the ways that we build and absorb history as narrative. Like any of the narratives we have played with using code in this class, we can explore the written documentation of history, to locate themes, word frequency, and more. Of course, an analysis of historical documents also means analyzing real-life social movements of the past, which reappear in current social issues.
On the paragraph beginning, "NLP is used to analyze text, allowing machines to understand..."
Reading about how machines can process and produce language based on the formulaic nature of a sentence made me wonder about linguistic relativity. The Sapir-Whorf hypothesis, which states that language and order / structure of language relates to how speakers organize their thoughts. It almost humanizes machines to think that the formulation of their linguistic function is so similar to that of our own.
There's a vaguely biblical structure to a creator building intelligence in its own form. Although I'm not religious myself, I think this is worth pursuing as there are some philosophical / literary implications that can inform how we approach the difficult moral questions about machine intelligence. To name a few, family / parenting / creating a new being, epiphany (or actualization of self-awareness), and purpose (What's the existential reason to construct a new consciousness?)
What gets counted counts
In data-centric PoliSci classes, there's oftentimes a dilemma around how to code gender. The short version is that variables in STATA are coded differently if they have a binary value (0 to 1) versus infinite options. When you mix gender into this, a dilemma arises.
The argument for making gender binary (0 to 1) is that the value for "Other / Refused" response is very low (In the Grinnell Poll from Oct. 2019, there were 17 O / R out of over 1000 respondents), and as a result has a massive confidence interval. As a result, it is considered unusable data, and thus inclusion is symbolic and unhelpful. Frankly, I think the bigger problem is that we lump "Other" and "Refused" into the same category - but if we separated them further, the confidence interval would get even larger.
The counterargument (in favor of coding with more than two options) is that not including the "Other / Refused is erasure; there are huge gaps in aid and help for transgender individuals that get so easily overlooked because of data paradigms described in the previous paragraph.
I guess this isn't so much a question as it is outlining a dilemma that I have no idea how to resolve. I pretty much always code gender with infinite possible responses myself, just because I feel like not drawing attention to the lack of usable data on "Other / Refused" isn't going to improve the situation.
Text is just a sequence of words, or more precisely, a sequence of characters. But when we usually deal with language modelling, or natural language processing, we are more concerned about the words as a whole, instead of just worrying about character-level depth of our text data.
I have two questions about this:
1) What would a 'missing value' be in language? 3) Not expecting an answer, I would like to reiterate: At what point does a machine become an agent responsible for its own actions? I know that solely sharing a processing structure (with different 'coding') isn't enough - but with more advanced AIs, the similarities only increase. Maybe we could answer that question with another question; at what developmental stage does a human become a responsible agent for their actions?
After ensuring that the wrangling did not introduce spurious information, Bari must therefore clean the data to get rid of the non-sensical data. In our class, you will sometimes clean data that is missing fields, such as a location that has a name, but not a latitude or longitude.
Missing values cause ambiguity problems in stats. Two sets of data might overlap because of the wider confidence intervals. At the same time, keeping missing values doesn't remedy that ambiguity (at least in PoliSci) because it just can't be used for anything; "refused" to answer a question can mean a lot of different things that the data scientist is unable to clarify.
Informal evidence of richer behavior:
Question of what "intelligence" actually is; is it an outward appearance / ability to answer questions posed, or does it require cogitation. On the road to actual generation of thought / human-ness, I'm curious about what point machines will be considered a moral agent unto themselves.
During World War II Konrad Zuse built the first program controlled digital computer that, instead ofBabbage's decimal arithmetic, used binary arithmetic implemented in on/ off electronics. This was a considerable simplification and made possible advances in increased speed and precision-important to our "digital" computers. Working independently (and very secretly) the British government cryptanaly-sis group of which Turing was part (and where he was instrumental in cracking the German Enigma code) created the Co/ossHs, which has been characterized as the first fully functioning electronic digital computer.4
Connection between technology / competition; computers were developed with the goal of gaining an edge in deriving information from Germany.
“Even though she’s growing up in India, she speaks and thinks in English
Relevant to our conversation about power and language; how technology provides an accessible platform for non-English languages to make a comeback.