[EI Seminar] Mohit Iyyer - Evaluating and Detecting Long-form LLM-generated Text

Speaker: Mohit Iyyer , UMass Amherst

Date: Thursday, November 16, 2023

Time: 4:00 PM to 5:00 PM Note: all times are in the Eastern Time Zone

Public: Yes

Location: 32-G575

Event Type: Seminar

Room Description:

Host: Athul Jacob, CSAIL MIT

Contact: Athul Paul Jacob, apjacob@mit.edu

Relevant URL:

Speaker URL: https://people.cs.umass.edu/~miyyer/

Speaker Photo:
None

Reminders to: seminars@csail.mit.edu

Reminder Subject: TALK: Mohit Iyyer - Evaluating and Detecting Long-form LLM-generated Text

Note: This event is *not* cancelled.

Abstract: Progress in NLP methodology over the last thirty years has been driven by benchmarks, from the Penn Treebank to GLUE. Benchmarks are useful because they provide a standard task, dataset, and means of evaluation that any researcher can use to quickly and easily demonstrate the value of their method. However, in the current age of LLMs, I argue that benchmarking is becoming increasingly obsolete. Beyond challenges such as data contamination, the dubious scientific validity of "prompt engineering", and usage of closed-source APIs, each of which is critical in its own right, there exist fundamental issues with how to formulate real-world tasks into benchmarks that can rank LLMs based on the much-desired "single score". I highlight these issues using some of my lab's recent work on tasks such as long-form question answering, book-length summarization, and literary translation. Next, I'll pivot to a different problem that plagues not only evaluation (e.g., via Mechanical Turkers using ChatGPT to complete tasks) but also society as a whole: the rapid proliferation of LLM-generated text. Detecting such text is not only important for combating malicious use cases such as academic plagiarism, but also to ensure that LLMs of the future are not just pretrained on text generated by their inferior predecessors. I outline several attacks against existing LLM-generated text detectors such as watermarking (e.g., paraphrasing, translation, cropping) and describe a retrieval-based approach that is more robust to these attacks but comes with issues of its own.

Bio: Mohit Iyyer is an associate professor in computer science at the University of Massachusetts Amherst, with a primary research interest in natural language processing. He is the recipient of best paper awards at NAACL (2016, 2018), an outstanding paper award at EACL 2023, and a best demo award at NeurIPS 2015, and he also received the 2022 Samsung AI Researcher of the Year award. He obtained his PhD in computer science from the University of Maryland, College Park in 2017 and spent the following year as a researcher at the Allen Institute for Artificial Intelligence.

Research Areas:
AI & Machine Learning, Human-Computer Interaction

Impact Areas:

This event is not part of a series.

Created by Athul Paul Jacob Email at Friday, November 10, 2023 at 3:10 PM.