Sign in

Lara J. Martin

@laramartin.net
447 followers 501 following 142 posts

Applying and improving #NLP/#AI but in a safe way. Teaching computers to tell stories, play D&D, and help people talk (accessibility). Assistant Prof in CSEE @ UMBC. 🏳️‍🌈♿ laramartin.net

PostsRepliesMedia
Reposted by Lara J. Martin
Jason Koebler @jasonkoebler.bsky.social · 22/09/2026
Scoop: Meta's Muse AI can make phone calls. Internally, Meta has told employees that many of these calls will be made by human beings in a call center: www.404media.co/meta-tests-m...
404media.co
Meta Tests Muse AI Agent Calls That Are Actually Made By Humans in a Call Center
"This has potential for so much negative PR. It could portray us as ‘their AI is not good enough so they still need humans’ kind of coverage for this launch."
26478180
Reposted by Lara J. Martin
Popular Science @popsci.com · 18/12/2025
The name "Popular Science" doesn't mean we shift our coverage depending on public opinion. It means we cover relevant subjects that are rigorously researched, reliable, and grounded in reality. And trans lives are grounded in reality. We see y'all. No matter what. www.popsci.com/science/tran...
popsci.com
First-of-a-kind study shows encouraging data for trans kids who socially transition
Ninety-four percent of participants in a new study stood firm in their trans identity after five years, and "detransitioning" is rare.
142176666710
Reposted by Lara J. Martin
Emily Halina @emilyhalina.bsky.social · 10/11/2025
Looking forward to seeing friends and colleagues in Edmonton this week for AIIDE! I'll be trying my hand at live tweeting the paper sessions at both the main conference and EXAG/INT, which I'll try to keep in this big reply thread (starting off tomorrow). @aiide.bsky.social @exag.bsky.social
1206
Lara J. Martin @laramartin.net · 11/11/2025
Shadab's Bluesky: @namer.bsky.social
000
Lara J. Martin @laramartin.net · 11/11/2025
We look at a common task in Augmentative and Alternative Communication (#AAC) where keywords are expanded into full sentences and see how using different emotion representations affects how LLMs generate sentences.
010
Lara J. Martin @laramartin.net · 11/11/2025
In this work, we introduce a new human evaluation of #LLMs called Representation Alignment. Representation Alignment looks at the similarity between how LLMs use conceptual representations and how humans interpret them.
Diagram of Representation Alignment evaluation. Four different emotion representations are passed to an LLM, along with keywords (e.g., wake, cat, morning). Each of these sentences are shown to the user, who has to pick the best-fitting sentence to the emotion representation they see.
130
Lara J. Martin @laramartin.net · 11/11/2025
I'm pleased to announce that Shadab Choudhury & Asha Kumar's paper "Evaluating Human-LLM Representation Alignment: A Case Study on Affective Sentence Generation for Augmentative and Alternative Communication" has been accepted to IJCNLP-AACL Findings 2025! arxiv.org/abs/2503.11881 #AACL2025 #IJCNLP
arxiv.org
Evaluating Human-LLM Representation Alignment: A Case Study on Affective Sentence Generation for Augmentative and Alternative Communication
Gaps arise between a language model's use of concepts and people's expectations. This gap is critical when LLMs generate text to help people communicate via Augmentative and Alternative Communication ...
221
Reposted by Lara J. Martin
Thomas Davidson @thomasdavidson.bsky.social · 09/10/2025
There is one week left to apply to join us at Rutgers! We're hiring an Assistant Professor in Computational Sociology as part of a cluster of new hires in data science and AI. Applications are due next Wednesday, 10/15.
jobs.rutgers.edu
Assistant Professor in Computational Sociology
The Department of Sociology at Rutgers University, New Brunswick, seeks applications for a tenure-track position at the Assistant Professor level specializing in Computational Sociology.  The search i...
01211
Lara J. Martin @laramartin.net · 30/09/2025
You can check out the pre-generated narratives based on Marvel movies here: what-if-game.github.io
what-if-game.github.io
What if...?
010
Lara J. Martin @laramartin.net · 30/09/2025
Runsheng (Anson) Huang's (who worked with me at Penn) paper "WHAT-IF: Exploring Branching Narratives by Meta-Prompting Large Language Models" is a system that turns existing plot summaries into engaging branching interactive fiction. Preprint: arxiv.org/abs/2412.10582
arxiv.org
WHAT-IF: Exploring Branching Narratives by Meta-Prompting Large Language Models
WHAT-IF -- Writing a Hero's Alternate Timeline through Interactive Fiction -- is a system that uses zero-shot meta-prompting to create branching narratives from a prewritten story. Played as an intera...
110
Lara J. Martin @laramartin.net · 30/09/2025
Patty Delafuente & Arya Honraopatil's paper "Does Reasoning Help LLM Agents Play Dungeons and Dragons? A Prompt Engineering Experiment" looks at using reasoning models like DeepSeek for generating actions of D&D players based on game state. Turns out this isn't something models can reason about!
121
Lara J. Martin @laramartin.net · 30/09/2025
Thrilled to announce that my lab has had two papers accepted to the Wordplay Workshop at EMNLP 2025!!
130
Lara J. Martin @laramartin.net · 29/08/2025
Congratulations to the winners of the Commonsense Personal Grounded Dialogue Challenge! discourse.aicrowd.com/t/winners-ca... we look forward to hearing about the details of your systems at the EMNLP 2025 Wordplay Workshop!
discourse.aicrowd.com
🏆 Winners & Call for Paper
Hello all, Thank you for being part of the Commonsense Persona Grounded Dialogue Challenge 2025! Whether you finished at the top or joined for the first time, we’re glad you participated in this seco...
031
Lara J. Martin @laramartin.net · 13/08/2025
Or maybe your dreams as a janitor will lead you to be the captain of your very own garbage ship!
Cover art for Sierra's old game. Space Quest V: Roger Wilco - The Next Mutation. Roger is holding a mop and surrounded by all sorts of aliens all standing waist-high in green goo
240
Reposted by Lara J. Martin
The Verge @theverge.com · 13/08/2025
Hidden Door is an AI storytelling game that actually makes sense
buff.ly
Hidden Door is an AI storytelling game that actually makes sense
You’ll help write what happens next.
4228
Lara J. Martin @laramartin.net · 13/08/2025
Submission deadline has been extended to September 12th! Everyone tell your friends! Submit your paper to this awesome workshop! wordplay-workshop.github.io
wordplay-workshop.github.io
/overview
Official website for the Wordplay Workshop at EMNLP 2025. Exploring interactive narratives, text-adventure games, and AI agents in language-based environments. Join us in Suzhou, China, November 5th-9...
074
Lara J. Martin @laramartin.net · 17/06/2025
The CFP for the Wordplay Workshop at #EMNLP2025 is out! We welcome any work that sits at the intersection of #NLP and games/narrative. We are looking for regular papers (4-8 pgs) and extended abstracts about open challenges in the space (2 pgs) Papers due: August 29 wordplay-workshop.github.io/cfp/
wordplay-workshop.github.io
/call_for_papers
Official website for the Wordplay Workshop at EMNLP 2025. Exploring interactive narratives, text-adventure games, and AI agents in language-based environments. Join us in Suzhou, China, November 5th-9...
053
Reposted by Lara J. Martin
Nature @nature.com · 02/05/2025
US President Donald Trump’s proposed budget for fiscal year 2026 calls for unprecedented cuts to scientific agencies that, if enacted, would deal a devastating blow to US science, policy specialists say. go.nature.com/4jYv3IP
go.nature.com
Trump proposes unprecedented budget cuts to US science
Huge reductions, if enacted, could have ‘catastrophic’ effects on US competitiveness and the scientific pipeline, critics say.
812269
Reposted by Lara J. Martin
👁️ @nathanjurgenson.com · 02/05/2025
every definition and description of "AI Slop" reads to me as what for-profit, highly metricized, and algorithmic platforms have always done. making it primarily an "AI" issue lets the platforms and this general logic too much off the hook imo
420163
Lara J. Martin @laramartin.net · 01/05/2025
What data is he pulling this from? because back when people were using Facebook, people would have 100+ "friends" on average. 😂
040
Reposted by Lara J. Martin
Dr. Casey Fiesler @cfiesler.bsky.social · 26/04/2025
This is one of the worst violations of research ethics I've ever seen. Manipulating people in online communities using deception, without consent, is not "low risk" and, as evidenced by the discourse in this Reddit post, resulted in harm. Great thread from Sarah, and I have additional thoughts. 🧵
251019462
Reposted by Lara J. Martin
Anthony Moser @anthonymoser.com · 26/04/2025
A key difference here is that while either can be incorrect, the structure of Wikipedia *creates context* and the structure of LLMs *destroys context* Wikipedia has linked sources and an edit history showing where information came from and who added it when An LLM just generates text
5740491213
Reposted by Lara J. Martin
Jeremy Berg @jeremymberg.bsky.social · 24/04/2025
ATTENTION: NSF GRANT RECIPIENTS We received a heads up from a trusted source that you should proactively download/print/screen shot any documentation on research.gov pertaining to your NSF awards, both those that are current and any that have closed in the last 5-6 years. 1/n
media.tenor.com
a cartoon of a man holding a frying pan and a spoon with red alert written above him
ALT: a cartoon of a man holding a frying pan and a spoon with red alert written above him
251038917
Reposted by Lara J. Martin
Anna Gifty @itsafronomics.blacksky.app · 19/04/2025
Northwestern just announced that the university will meet the funding needs of any research that is being impacted by stop orders. LET'S GOOOOOOOOOOOOO! #edusky #academicsky wgntv.com/evanston/nor...
wgntv.com
Northwestern University to self-fund research despite Trump administration freeze
While university officials said they’ve still not received official word of such action, they acknowledged receiving about 100 stop-work orders from the federal government on roughly 100 fede…
142125608
Reposted by Lara J. Martin
Amy J. Ko @amyko.phd · 19/04/2025
My NSF CS for All grant was just terminated (www.nsf.gov/awardsearch/...). Nationwide, we were: • Improving the quantity + quality of K-12 CS teachers • Ensuring the sustainability of K-12 CS teacher prep programs • Ensuring all youth in the U.S. have access to great CS teachers
nsf.gov
NSF Award Search: Award # 2318257 - Collaborative Research: An Equitable, Justice-Focused Ecosystem for Pacific Northwest Secondary CS Teaching Lock
28283108
Reposted by Lara J. Martin
Lisa Fazio @lkfazio.bsky.social · 18/04/2025
Can confirm that my NSF grant "How False Beliefs Form & How to Correct Them" was cancelled today because it is "not in alignment with current NSF priorities" Shocking that understanding how people are misled by false information is now a forbidden topic. Our work will continue but at a smaller scale
14230681363
Reposted by Lara J. Martin
Felipe Pepe @felipepepe.bsky.social · 18/04/2025
I love how you can have an overview of the internet in 1995 by... looking in a book that lists 8,000+ websites with photos lol
411222
Lara J. Martin @laramartin.net · 14/04/2025
I really love this idea that venture capitalists and tech companies are just doing reinforcement learning IRL: "With access to infinite resources, these adherents ran reinforcement learning on their companies. They threw billions of dollars' worth of spaghetti at the wall until something stuck."
030
Reposted by Lara J. Martin
Janelle Shane @janelleshane.com · 13/04/2025
1. LLM-generated code tries to run code from online software packages. Which is normal but 2. The packages don’t exist. Which would normally cause an error but 3. Nefarious people have made malware under the package names that LLMs make up most often. So 4. Now the LLM code points to malware.
11978743580
Reposted by Lara J. Martin
Jonathan Cohn @jonathancohn.bsky.social · 13/04/2025
The bipartisan belief that America would be a better place if *other people* worked in a factory. www.ft.com/content/8459...
80+% of both Democrats and Republicans believe America would be better off if more Americans worked in manufacturing, but < 30% say that about themselves.
23291252809
Reposted by Lara J. Martin
Liz Fong-Jones (方禮真) @lizthegrey.com · 11/04/2025
I am *horrified* by Sam Altman's suggestion that ethical AI limits can be gathered through AI itself being used to scale out having conversations with hundreds of millions of human users of OpenAI products and it being a matter of majority consensus mediated through his machine. Ethics is a field!
1127251
Lara J. Martin @laramartin.net · 10/04/2025
I literally thought it was an Onion article because of that
0190
Reposted by Lara J. Martin
Margot Finn @eicathomefinn.bsky.social · 09/04/2025
'Bluesky has overtaken its flailing rival X in hosting posts related to new academic research, indicating the platform is fast becoming the go-to place for scholars to share their work.'
timeshighereducation.com
X’s dominance ‘over’ as Bluesky becomes new hub for research
Data indicates more scholars turning to alternative social media site to post about their work after Elon Musk’s Twitter takeover
129174324387
Reposted by Lara J. Martin
Thinking Person's Guide to Autism @thinkingautism.com · 07/04/2025
Reminder that for two more days @assistiveware.bsky.social is offering a 50% discount on your favorite AssistiveWare apps like the symbol/text-to-speech of Proloquo2Go. Sale ends April 9. www.assistiveware.com/aac-app-sale #AutismAcceptanceMonth
assistiveware.com
Autism Acceptance Month Sale 2025 - AssistiveWare
Everyone has something to say. Working closely with the AAC community, we build apps and share best practice to help you communicate with the world.
1118
Reposted by Lara J. Martin
Michael Yarbrough (he/him) @mwyarbrough.bsky.social · 07/04/2025
The resolution is on the desk of Rutgers President Jonathan Holloway as we speak. If you think he should move forward with it, drop him a note at president at rutgers dot edu to say so. Anyone with an interest should write! And that goes double for Rutgers alums + affiliates!!
1279105
Lara J. Martin @laramartin.net · 06/04/2025
It's almost as if the US values sports more than art...
000
Reposted by Lara J. Martin
Jason Koebler @jasonkoebler.bsky.social · 04/04/2025
Petabytes of critical government research that is marked for possible deletion cannot be archived because of technical and legal restrictions, meaning it is very likely to be lost forever: www.404media.co/nih-archives...
404media.co
Massive, Unarchivable Datasets of Cancer, Covid, and Alzheimer's Research Could Be Lost Forever
Days before Robert F. Kennedy Jr. announced that 10,000 HHS staffers would lose their jobs, a message appeared on NIH research repository sites saying they were "under review."
18843482758
Reposted by Lara J. Martin
Kendra "Gloom is My Beat" Pierre-Louis @kendrawrites.com · 05/04/2025
Almost universally the handful of factories that remain in the US that make clothes rely on immigrant labor which we're deporting. And making clothes is a *skill*. There's no shortcuts beyond hours behind a machine.
950252
Lara J. Martin @laramartin.net · 03/04/2025
Ooo show them the cross stitch I made!! 😂
010
Lara J. Martin @laramartin.net · 03/04/2025
New Jersey is just stepping up left and right
040
Reposted by Lara J. Martin
Scott Delaney @scott-delaney.bsky.social · 02/04/2025
If your NIH grant has been terminated, please report it here 👇 The more information we collect, the more we can organize, advocate, and fight back! forms.gle/81iLikdBtmCP...
forms.gle
Cancelled NIH grant information submission form
Please use this form to submit information identifying specific NIH grants that have been cancelled for any reason after January 20, 2025. Data from this form will be used to update the Terminated NI...
310680
Reposted by Lara J. Martin
Martin Enserink @martinenserink.bsky.social · 01/04/2025
This hasn't gotten much attention—but new National Science Foundation grants have dropped by almost half since Trump took office, an analysis by @davidimiller.bsky.social shows. www.science.org/content/arti... (Story by @policyhound.bsky.social)
science.org
NSF has awarded almost 50% fewer grants since Trump took office
The reasons aren’t clear—and the agency’s director claims the president’s policies haven’t slowed grant awards
610285
Reposted by Lara J. Martin
Brosephine Wires, Perennial Soup-Lover @joparkerbear.bsky.social · 01/04/2025
This is scary as shit. The part about Indiana University erasing his existence entirely from their system is particularly chilling. www.wired.com/story/cybers...
wired.com
Cybersecurity Professor Mysteriously Disappears as FBI Raids His Homes
Xiaofeng Wang, a longtime computer science professor at Indiana University, has disappeared along with his wife, and their profiles on the school's website were wiped ahead of recent FBI raids.
9481265
Lara J. Martin @laramartin.net · 01/04/2025
Happy Trans Day of Visibility to all the lovely trans women, trans men, and everyone else within and without the gender spectrum. ♥️ #tdov
030
Reposted by Lara J. Martin
Dare Obasanjo @carnage4life.bsky.social · 29/03/2025
This is going to be the vibe coding version of the Chernobyl disaster, a wholly preventable failure caused by a combination of hubris, incompetence and design flaws.
67389942644
Lara J. Martin @laramartin.net · 30/03/2025
Research is only niche for people lacking imagination.
000
Reposted by Lara J. Martin
VGBees 🎮🐝🐝🐝 @vgbees.com · 27/03/2025
The Elden Ring difficulty discourse betrays disabled gamers in important ways
vgbees.com
How the Elden Ring Difficulty Discussion Fails Disabled Gamers - VGBees
Everyone is rightfully sick of the FromSoftware difficulty conversation, most of all me, but I think there’s a perspective here we’re not exploring as thoroughly...
02511
Lara J. Martin @laramartin.net · 27/03/2025
These solutions aren't perfect, but I think it shows that people are starting to care. We're starting to research alternatives. We're starting to listen to people. So, if I had to simplify my point of view, I can sum it up as pro-research, but anti-venture capitalist. 4/4
020
Lara J. Martin @laramartin.net · 27/03/2025
There are some solutions to making GenAI more customizable (by using neurosymbolic methods), use less copyrighted material (like models like OLMo), or be less harmful to the environment (training smaller models or quantized models so the bigger models don't have to be used over and over again). 3/
110