Alex Choi
@alex_choi_cs
Followers
38
Following
69
Media
7
Statuses
23
student @GeorgeMasonU. NLP research with GMNLP. always curious.
Joined October 2023
🧑🤖The LLM Effect: Are Humans Truly Using LLMs, or Are They Being Influenced By Them Instead❓ This is obviously a massive question that is both important and timely. Accepted to #EMNLP2024 Main, our paper examines the integration of LLMs into specialized expert workflows, and
2
5
14
🌴 We'll be presenting our work at #EMNLP2024 Wednesday (Nov 13th) from 10:30am-12:00pm in Riverfront Hall!! Stop by to hear about it and discuss!👋 @emnlpmeeting w/ @SyedaSabrina11 @anas_ant
0
0
3
When is using grammars for producing translation systems beneficial? In our work with Jonathan Hus we answer this question. Find us tomorrow Tue at 4pm by our poster! Paper: https://t.co/skleP6v5k6 1/2
1
9
32
On my way to MIAMI, to attend EMNLP! We will be presenting our paper on the poster session D. If you are attending and want to catch up, DM me. Also, feel free to drop by and talk NLP.
🧑🤖The LLM Effect: Are Humans Truly Using LLMs, or Are They Being Influenced By Them Instead❓ This is obviously a massive question that is both important and timely. Accepted to #EMNLP2024 Main, our paper examines the integration of LLMs into specialized expert workflows, and
0
1
5
search engines like google cuts off the serendipity discovery allowed by library shelves / google maps cuts off the user's spatial understanding of their surrounding / automatic differentiation of deep nets cuts off your deep understanding of gradient flows / calculators cuts ...
Setting things up so that you get "the answer" to your question cuts off the user's ability to do the sense-making that is critical to information literacy. >>
3
6
140
As OpenAI and Meta introduce LLM-driven searchbots, I'd like to once again remind people that neither LLMs nor chatbots are good technology for information access. A thread, with links: >>
117
1K
5K
Check out our award-winning work from #SIGSPATIAL24! Come for the synthetic travel survey data generation, stay for the thorough evaluation on this super tricky setting! Details in 🧵👇
Travel survey data are vital for urban mobility assessments—but they’re often costly and difficult to collect. Could LLMs help us synthesize such data? Our latest work, in collaboration with @anas_ant and @dpfoser, and the Best Paper Winner🥇at #SIGSPATIAL24, shows they can! 🧵👇
0
4
16
Have you ever thought of trajectories as sentences? See the figure, it's not too weird a thought! Turns out we can use language modeling ideas to do trajectory modeling! Paper ( https://t.co/SnaGnNjZjy) to be presented later today at ACM SIGSPATIAL. See 🧵for more details 👇
4
7
44
Thank you to those who participated in our recent all-member vote regarding our name change. The change is happening! We are: The Nations of the Americas Chapter of the Association for Computational Linguistics! Announcement 👉
0
13
65
The scientific and societal conversation surrounding LLMs is massive and we hope that our study can add meaningful evidence to this ongoing global dialogue. We will be presenting in person at #EMNLP2024 in November! Please stop by! preprint available now on arXiv.
arxiv.org
Large Language Models (LLMs) have shown capabilities close to human performance in various analytical tasks, leading researchers to use them for time and labor-intensive analyses. However, their...
0
0
1
As well as introduce an anchoring bias, thus influencing the expert decision making process. 5/6
1
0
1
But on the other hand, LLM suggestions can come with real costs. When tasked with discovering latent concepts within documents, LLMs can miss low prevalence, high importance topics. 4/6
1
0
0
Our results indicate both positive efficiency gains, but also the possibility of worrisome emerging trends. On the one hand, when it came to the task of topic assignment, our experts were able to operate much more quickly with LLM suggestions. 3/6
1
0
0
In our multi-stage study, we asked experts in the domain of International Policy to analyze and annotate lengthy AI Policy interviews, through the framework of Topic Modeling. In Stage 1 they discovered latent topics within one set of lengthy documents, and in Stage 2 they
1
0
0
We have seen the usual man:doctor🧑⚕️ and woman:nurse👩⚕️ stereotypes. But guess what? VLMs are also throwing out wild associations like blonde:dumb🤦♀️, old person:dinosaur🦕, or college student:broke. 🧑🎓💸 BiasDora, our #EMNLP2024 Findings Paper: https://t.co/5UgT4irJs9, explores
1
6
13
Congratulations to dr. Mahfuz ibn Alam, the first PhD our of my lab, who defended his PhD yesterday! 🎉 Mahfuz has contributed to Machine Translation for under-served varieties, starting with translation of non-native English (best paper at WNUT'20 -- published before we ... 1/3
11
5
105
What happens when students in an intro programming class get access to GPT-4? A trade-off: “positive benefits on exam performance for adopters” but “an average decrease in participation in several class elements” for most but an increase for those from less-developed countries.
6
36
237