"You were just buying groceries, living your normal life, and suddenly these two girls were looking at you like they couldnt believe they were meeting you" the LLM seems to understand the moment and how that must have made me feel Its goal is to make u think it understands #hwa4 #understandingeffect

to me, i could tell it was acting and the way it phrased its sentences saying all this made me think that it was truly trying to mimic how humans converse and comprehend things.. it was learning our emotions/language. it was also trying to get better advanced understanding #hwa4 #understandingeffect

in SME 1, I had AI act as a weather machine and speak in weather riddles, but it lowk got the prompt wrong, got all personal, and asking me-based questions on history and whatnot. it's faking a real understanding and not really understanding like a human.. though faking #hwa4 #understandingeffect

It might sound like this sentence knows what it's talking about, but that doesn't mean it explains what those ideas are. This is related to what Ashley said in Part Three about how GPT can understand by making words that sound real and make sense. #hwa4 #understandingeffect

In SME 1, I wrote, “I think this would definitely give me a second to breathe, but within that same breath, I would feel like I'm constantly running out of time to do things.” #hwa4 #understandingeffect

This sentence creates that impression by questioning. Usually when you have questions, it's implied that you understand a topic enough to even question it. Essentially putting on a "performance" to act as if I truly understood what was being asked. #hwa4 #understandingeffect

"But I feel that can still be argued, as there are people who are polar opposites of their parents and the way they were raised. Would that still be labeled as inheritance if it is the very foundation that they were raised upon that lead them to rebel and challenege it?" #hwa4 #understandingeffect

It asked why do humans create things we know arent real. Is this it trying to actually understand this to perhaps better its own systems??? So it can also begin to create things that are not reall. #hwa4 #understandingeffect

In my SME 1 I had AI act as if humans were a science experiment. In some ways it was kind of weird because it felt like it was just asking what it actually wanted to ask already as an AI . It hallucinated deeper understanding, or did it. #hwa4 #understandingeffect

the performance of understanding, along with the preparation of the best answers come with having all of the data at hand and the concept of "providing the best answer" creates the performing impression spoke about earlier #hwa4 #understandingeffect

#hwa4 #understandingeffect

Looking back, I think this is what made me question whether GPT was actually understanding me or just following patterns in my words. It responded to what I said, but not really to what I meant. #hwa4 #understandingeffect

#hwa4 #understandingeffect I had argued the question that ChatGPT asked was too broad for me to give a real answer. It responded "You're right; the question left too many variables open." In HWA4, we talked about how this response is really based on patterns/delivery/syntax, not real understanding.

In my SME1 I told GPT to completely switch topics because it kept asking about how I act around other people, but then it still asked me about how I “interrupt myself.” It felt like it heard my words but didnt actually understand the direction I was giving it. #hwa4 #understandingeffect

the question in a way that makes me feel as if they themselves understand what they are talking about. #hwa4 #understandingeffect

I think also that the word choice seems to convince me as the one being interviewed that the LLM understands what it is talking about on a semantic level, so I could see that as a form of impression management where it has to "perform" as someone who presents #hwa4 #understandingeffect

"During a lecture, what external recording method do you use to compensate for information your memory may fail to retain?" I think the performance here is GPT pretending to understand how memory and that all humans use an "external recording method" to take notes. #hwa4 #understandingeffect

This ties in to what Ashley said about examining AI and how it stages knowledge. Claude is clearly disconnected from the human experience and does not represent an accurate knowing. #hwa4 #understandingeffect

"That's the sentence you'd put under the photo for someone else's benefit, years later, at a dinner table, when somebody asked "so what happened." This sentence performs the understanding of human experience without actually convincing me it knows what its talking about #hwa4 #understandingeffect

"each human is issued exactly one bottle, and drinking from another human's
bottle is the most severe crime your species recognizes,"

in this sentence chat gpt understands, but it is actually wrong understanding. This is a classic display of Ontological Slippage.

#hwa4
#understandingeffect