-
MSHLT Student Charlie Accurso
Introducing MSHLT student Charlie Accurso
-
MSHLT Student Ashwin Raj
Introducing MSHLT student Ashwin Raj
-
Who and What comprise AI Skepticism? | Eric Jackson
A re-post of a re-post? Yes, I'm sorry, but I'm going to do it anyway because I found this essay helpful (and found it because of Gary Marcus). I like the way that this essay lays out the different kinds of "AI Skeptics" there are--and I think that these are possibly not mutually-exclusive labels, a fact which I think the author of this post, Benjamin Riley , would agree with. The labels are helpful to summarize some of the nuances of the perspectives on AI that people hold, and although some of the labels denote different nuances, some of them can be held at the same time by the same person. I found it helpful to reflect on where in this taxonomy my own perspective would land. Large language models are doing 𝘴𝘰𝘮𝘦𝘵𝘩𝘪𝘯𝘨, and sometimes that something can be helpful, but if you don't already know enough about the problem you're asking it to solve, it can also lead you astray (as I've had to demonstrate for my freshman formal methods class many times this semester), and it does come with serious 𝘤𝘰𝘴𝘵𝘴. A worthwhile read! Thank you for this take on synthesizing the "AI Skeptics" field, Benjamin!
-
Vector Databases Are the Wrong Abstraction | Eric Jackson
This morning, I had just finished preparing a video+code module that introduces Retrieval-Augmented Generation systems to students in my grad-level 𝘐𝘯𝘧𝘰𝘳𝘮𝘢𝘵𝘪𝘰𝘯 𝘙𝘦𝘵𝘳𝘪𝘦𝘷𝘢𝘭 course. Then I saw this post, thanks to a comment by Keegan Reeve. This article includes a lot of helpful observations (and improvements to implementing a database of vector embeddings) for the IR-portion of a RAG system. I hope my students are listening! Thanks, Daniel Ruiz Riquelme for highlighting this post from Timescale!
-
How language data can benefit your organization | Eric Jackson posted on the topic | LinkedIn
Your data enables insights. Your data drives decision-making. Understanding your data brings benefits, to you and your organization. 𝗪𝗵𝗮𝘁 𝗶𝗳 𝘆𝗼𝘂𝗿 𝗱𝗮𝘁𝗮 𝗶𝗻𝘃𝗼𝗹𝘃𝗲𝘀 𝗵𝘂𝗺𝗮𝗻 𝗹𝗮𝗻𝗴𝘂𝗮𝗴𝗲? What do I mean by data that involves human language? I mean passages of text, short or long; responses to a single question, or technical documents of hundreds of pages. This data might be structured or semi-structured, perhaps in a relational database or in a collection of submissions from a web form. It might be data that is completely unstructured, perhaps in a large collection of text files or PDFs. • 𝗜𝘀 𝗶𝘁 𝘁𝗼𝗼 𝗹𝗮𝗿𝗴𝗲? Are there decisions you need to make based on that data, but you have too much data to analyze easily by having a person look over it? • 𝗜𝘀 𝗶𝘁 𝘁𝗼𝗼 𝗰𝗼𝗺𝗽𝗹𝗲𝘅? Is there information that you need to extract from that data? Do you need to answer questions based on that data? Summarize part or all of that data? Maybe even create a search tool to answer arbitrary questions based on that data? • 𝗔𝗿𝗲 𝘆𝗼𝘂 𝘀𝗵𝗼𝗿𝘁-𝘀𝘁𝗮𝗳𝗳𝗲𝗱? Are you or your organization missing out on benefitting from language data because you or your team lack the skills to work with it? Or maybe you have people with the skills to analyze the data, but they don't have enough time to address something that you see as a strategic problem. Students in our 𝘔𝘢𝘴𝘵𝘦𝘳'𝘴 𝘪𝘯 𝘏𝘶𝘮𝘢𝘯 𝘓𝘢𝘯𝘨𝘶𝘢𝘨𝘦 𝘛𝘦𝘤𝘩𝘯𝘰𝘭𝘰𝘨𝘺 program are trained in programming and NLP skills, and each of them needs to complete a practical project where they apply these skills in a real-world task. 𝗦𝗼𝗹𝘃𝗶𝗻𝗴 𝘆𝗼𝘂𝗿 𝗽𝗿𝗼𝗯𝗹𝗲𝗺 𝗰𝗼𝘂𝗹𝗱 𝗯𝗲 𝘁𝗵𝗲 𝗵𝗶𝗴𝗵𝗹𝗶𝗴𝗵𝘁 𝗼𝗳 𝘁𝗵𝗲𝗶𝗿 𝘀𝘁𝘂𝗱𝘆 𝗽𝗿𝗼𝗴𝗿𝗮𝗺! Got language data? Let's start a conversation. Working with one of our HLT students might bring you the solution you're looking for!