Griffiths’ paper urges for a change in the field of cognitive behavior to not only use computers as an assistive tool in cognitive research, but to implement computers altogether as an aspect of cognitive science to study. The sorts of changes Griffiths describes aptly compare to current trends of using extremely large corpuses and feeding […]
Continue reading →Hello everyone, Firstly, a big round of applause to both teams for their commendable efforts in the debate! The depth and sophistication of the discussions were enlightening, and I personally also gained many new insights from your exchanges and performances. Thank you! The debate was a great commencement to the latter half of our course. […]
Continue reading →– Introduction of knowledge: For analysis; For utilization; Outside of domain – “Chain-of-thought” reasoning – Counterfactuals/hypotheticals – Recursive phrasing – Constraints/lists – Adjusting prompt phrasing to test the result – Ethical scenarios – Language – Cherry-picking – Logic – Translation – Summarization – Training the model to ignore irrelevant deviations (e.g. typo) – Prompts pp. […]
Continue reading →Objective of the Debate The primary objective of the debate is to foster open discussion and collaborative thinking, rather than focusing on competition or a win-lose outcome. While participants will adopt and defend particular positions, the ultimate goal is to explore and comprehend the rationale behind differing reasoning and perspectives. Focus on Clarity, Not […]
Continue reading →I agree with Roose’s point in “How Schools Can Survive (and Maybe Even Thrive) With A.I. This Fall”, that generative AI should instead be understood and not banned in classrooms. I think that for many people, especially those who are not experienced in working with AI or do not know much about how models like […]
Continue reading →Hey everyone, We’re now stepping into the second half of our course, focusing much more on evaluating LLMs and their real-world impacts. We’re moving into territory where there aren’t always clear answers, and I’ll increasingly count on your interests and contributions to help steer our discussions, less lecturing, more chatting and debating. For Monday: We’re […]
Continue reading →Some aspects of Andrew Abbott’s critique resonate with me. For instance, he points out that some students only read the highlighted portions of the text, bypassing the development of a logical argument that comes with reading the entire passage. Having attended his class for two weeks, I confess to being one of those students who […]
Continue reading →Dr. Chomsky’s argument against LLMs primarily focuses on how despite the perceived intelligence of a model, any given is unable to process information the way humans can, which, in his opinion, poses a limit on LLMs. I agree that the way LLMs are constructed limit their ability to imitate humans, however I think Chomsky is […]
Continue reading →Chomsky’s counterargument, in my view, lacks clarity. He contends that “The essence of machine learning is about description and prediction; it doesn’t delve into causal mechanisms or physical laws.” Yet, when we prompt advanced models like GPT 3.5 or GPT 4 with questions such as why apples fall, they can provide explanations that touch on […]
Continue reading →From Week 4, we’ll start our Monday class for an about 40 minutes lecture finishing the two approaches to language models, and after that I’ll talk about why and how to make language models large. Then we’ll spend the second half of Monday class and the whole Wednesday class to deepen our exploration into the behaviors, […]
Continue reading →