Reflecting on Language Technologies, AI, and our Society [50 points]
For this last homework, we will do something entirely different: reading articles and writing a response essay. We will also cover one last important topic: the current status of computational linguistics, NLP, and AI.
Having spent the semester focused on acquiring foundational knowledge, now is a good time to step back and reflect on the larger picture. If you haven't heard, Artificial Intelligence is transforming our society and the world, and language technologies are at the front and center. Incidentally, there are tons of fascinating books and in-depth reports being written about this topic: they are intended for the curious general public, but nevertheless provide a wealth of technical details and insights for us fledgling computational linguists.
I know you're all thinking ChatGPT and LLMs, but we have to start with the beginning of this revolution, which was kick-started with the development of deep learning and neural net technologies. This NYT article from 2016 presents a close-up view:
Fast forward to 2022, ChatGPT and LLMs (Large Language Models) burst into the scene and completely changed the landscape. First, we must understand how they work. These two articles do a marvelous job of unpacking their technical underpinnings:
By now, you have seen plenty of media coverage on how LLMs and Generative AI are taking over the world. But what does this mean for linguists, the field of NLP, and our society at large? First, an article featuring Emily Bender, a computational linguist emerging as a leading voice, and then, a survey/position article hot off the press on the impact of LLMs and the future of NLP.
Preparation:
- Some of the readings are LONG! Make sure to set aside enough time. They will be anything but boring though -- I promise!
- "The Great A.I. Awakening" in particular is super long. Feel free to skim through some sections focusing on older historical background. Don't be put off by the length though -- many students have told me this one ended up being their favorite.
- For NYT, you can create a free-access account using your Pitt email. Some news outlets implement a pay wall. I recommend saving the web pages as a PDF document on your computer.
Your essay, what to write, and how:
- 800 words minimum.
- Don't just summarize the articles! Your personal impression and perspective are wanted.
- Integrate the knowledge you learned in class, including the lecture content and readings from Jurafsky & Martin and also Language and Computers.
- Also, the technologies prominently featured in the reading (Deep Learning, neural networks, Google Translate, BERT, GPT-3, ...) are the absolute cutting-edge, industrial-grade NLP technologies that go much beyond what we can properly learn in the classroom in one short semester. But learning about them is part of the goals of this homework.
- Which means: some of the content will be in the final exam.
- One last thing. Reading the articles, you may notice the presence of tension between voices from NLP scholars and those from journalists. We shouldn't automatically dismiss publications written by non-academics; these articles are all clearly backed by extensive research. However, know that ultimately the authors are reporters who do not have deep expertise and also are writing for a general audience in the context of an intense societal hype around AI. That is to say: you should keep a critical eye.
SUBMIT:
- A word-processed file. MS Word or PDF format.
|
|
|