Princeton Laboratory for Artificial Intelligence
@princetonainews.bsky.social
Supporting and expanding the scope of AI research at Princeton.
Reposted by Princeton Laboratory for Artificial Intelligence
New research from Princeton and the University of Chicago shows that AI can develop prejudices in the hiring process faster than humans might.
AI can invent its own hiring bias — and it happens fast, new study finds
Researchers found AI models stereotyped job candidates more than humans did. A Phoenix jobs expert says AI interviews are already common here.
www.azfamily.com
July 30, 2026 at 2:30 PM
New research from Princeton and the University of Chicago shows that AI can develop prejudices in the hiring process faster than humans might.
Reposted by Princeton Laboratory for Artificial Intelligence
What do rubber ducks have to do with learning and AI? You can find out in my TEDx talk, released today as an Editor’s Pick at the YouTube TED Channel!
www.youtube.com/watch?v=cbiy...
www.youtube.com/watch?v=cbiy...
How you can learn in a world of information overload | Tania Lombrozo | TEDxNewEngland
YouTube video by TEDx Talks
www.youtube.com
July 27, 2026 at 5:14 PM
What do rubber ducks have to do with learning and AI? You can find out in my TEDx talk, released today as an Editor’s Pick at the YouTube TED Channel!
www.youtube.com/watch?v=cbiy...
www.youtube.com/watch?v=cbiy...
Reposted by Princeton Laboratory for Artificial Intelligence
🚨 New @iclr_conf paper! 🚨
Learning is Forgetting: LLM Training As Lossy Compression
by @henryconklin.bsky.social, @tomhosking.bsky.social, Yi-Chern Tan, Julian Gold, Jonathan Cohen, @cocoscilab.bsky.social, @maxbartolo.bsky.social and @seraphinagt.bsky.social
Arxiv: arxiv.org/abs/2604.075... 🧵
Learning is Forgetting: LLM Training As Lossy Compression
by @henryconklin.bsky.social, @tomhosking.bsky.social, Yi-Chern Tan, Julian Gold, Jonathan Cohen, @cocoscilab.bsky.social, @maxbartolo.bsky.social and @seraphinagt.bsky.social
Arxiv: arxiv.org/abs/2604.075... 🧵
April 14, 2026 at 11:33 AM
🚨 New @iclr_conf paper! 🚨
Learning is Forgetting: LLM Training As Lossy Compression
by @henryconklin.bsky.social, @tomhosking.bsky.social, Yi-Chern Tan, Julian Gold, Jonathan Cohen, @cocoscilab.bsky.social, @maxbartolo.bsky.social and @seraphinagt.bsky.social
Arxiv: arxiv.org/abs/2604.075... 🧵
Learning is Forgetting: LLM Training As Lossy Compression
by @henryconklin.bsky.social, @tomhosking.bsky.social, Yi-Chern Tan, Julian Gold, Jonathan Cohen, @cocoscilab.bsky.social, @maxbartolo.bsky.social and @seraphinagt.bsky.social
Arxiv: arxiv.org/abs/2604.075... 🧵
Language models’ ability to understand 3D space is improving, but they still lag behind in spatial reasoning.
A new benchmark from AI Lab researchers uses the Rubik’s Cube to evaluate how well language models handle spatial transformation over time.
bit.ly/4sGnBGx
A new benchmark from AI Lab researchers uses the Rubik’s Cube to evaluate how well language models handle spatial transformation over time.
bit.ly/4sGnBGx
Evaluation inspired by Rubik’s Cube tests language models’ spatial reasoning abilities
In the workplace, AI agents and other language models have become useful tools for tasks like writing code or navigating websites. However, researchers have hit significant obstacles in translating la...
bit.ly
April 8, 2026 at 7:47 PM
Language models’ ability to understand 3D space is improving, but they still lag behind in spatial reasoning.
A new benchmark from AI Lab researchers uses the Rubik’s Cube to evaluate how well language models handle spatial transformation over time.
bit.ly/4sGnBGx
A new benchmark from AI Lab researchers uses the Rubik’s Cube to evaluate how well language models handle spatial transformation over time.
bit.ly/4sGnBGx
Reposted by Princeton Laboratory for Artificial Intelligence
This week, @princeton.edu is hosting the 29th Annual ALTA Conference, April 9–11, co-organized by @happybuzaaba1.bsky.social, Associate Research Scholar at @princetonainews.bsky.social and CDH affiliate! Theme: "Reimagining African Language Pedagogy in the Digital Age." Open to all to attend.
ALTA 2026: Reimagining African Language Pedagogy In the Digital Age
ALTA 2026: Reimagining African Language Pedagogy In the Digital Age
cdh.princeton.edu
April 6, 2026 at 8:16 PM
This week, @princeton.edu is hosting the 29th Annual ALTA Conference, April 9–11, co-organized by @happybuzaaba1.bsky.social, Associate Research Scholar at @princetonainews.bsky.social and CDH affiliate! Theme: "Reimagining African Language Pedagogy in the Digital Age." Open to all to attend.
On Mar. 18, Karthik Narasimhan sat down with Ankush Gola ’15 to discuss @langchain.bsky.social, the startup he co-founded that provides the framework for building AI agents and LLM-powered applications.
ai.princeton.edu/news/2026/la...
ai.princeton.edu/news/2026/la...
LangChain co-founder, CTO comes to campus for Founders Talk Series
On Mar. 18, Karthik Narasimhan, associate professor of computer science, sat down with Ankush Gola ‘15 to discuss LangChain, the startup he co-founded that provides the framework for building AI agent...
ai.princeton.edu
April 2, 2026 at 3:15 PM
On Mar. 18, Karthik Narasimhan sat down with Ankush Gola ’15 to discuss @langchain.bsky.social, the startup he co-founded that provides the framework for building AI agents and LLM-powered applications.
ai.princeton.edu/news/2026/la...
ai.princeton.edu/news/2026/la...
What if every time you visited a page online you could expect something new and personalized?
Enter the Web World Model, a framework that maintains the logical consistency of standard web code while enabling the unlimited environment of an LLM: bit.ly/4aMwIzF
Enter the Web World Model, a framework that maintains the logical consistency of standard web code while enabling the unlimited environment of an LLM: bit.ly/4aMwIzF
Web World Model creates unlimited website environments without sacrificing logical consistency
What if every time you visited a page online you could expect something entirely new and personalized?“Existing websites are boring,” said Shilong Liu, a postdoctoral research fellow for AI for Accele...
bit.ly
March 3, 2026 at 4:07 PM
What if every time you visited a page online you could expect something new and personalized?
Enter the Web World Model, a framework that maintains the logical consistency of standard web code while enabling the unlimited environment of an LLM: bit.ly/4aMwIzF
Enter the Web World Model, a framework that maintains the logical consistency of standard web code while enabling the unlimited environment of an LLM: bit.ly/4aMwIzF
Reposted by Princeton Laboratory for Artificial Intelligence
Next week! For our Modeling Culture program, @zoeleblanc.bsky.social presents two projects exploring how computational methods can help us understand past and present struggles to make information more just. Join us on Tuesday, March 3, 4:30 PM @ CDH!
Liberating Information: Exploring Computational Methods to Study Histories of Information, from Decolonization to Digital Humanities
Liberating Information: Exploring Computational Methods to Study Histories of Information, from Decolonization to Digital Humanities
cdh.princeton.edu
February 25, 2026 at 7:19 PM
Next week! For our Modeling Culture program, @zoeleblanc.bsky.social presents two projects exploring how computational methods can help us understand past and present struggles to make information more just. Join us on Tuesday, March 3, 4:30 PM @ CDH!
Reposted by Princeton Laboratory for Artificial Intelligence
Princeton is leading a new wave of collaboration, from using LLMs to piece together ancient fragments to humanists helping engineers build more ethical, accessible models. This is the future of interdisciplinary research. #DigitalHumanities #History #AI #Princeton
AI and the humanities: Across the Princeton campus, an era of collaboration is underway.
Princeton has firmly established its presence at the forefront of AI research — including transformative work in humanities scholarship.
www.princeton.edu
February 24, 2026 at 11:12 PM
Princeton is leading a new wave of collaboration, from using LLMs to piece together ancient fragments to humanists helping engineers build more ethical, accessible models. This is the future of interdisciplinary research. #DigitalHumanities #History #AI #Princeton
New Deep Dive from @anilananth.bsky.social on our blog:
@msalganik.bsky.social & colleagues developed a way to create and analyze an individual’s "Book of Life." The eventual goal? Train language models on data from millions of Books of Life to predict an individual's life outcome: bit.ly/473NmIy
@msalganik.bsky.social & colleagues developed a way to create and analyze an individual’s "Book of Life." The eventual goal? Train language models on data from millions of Books of Life to predict an individual's life outcome: bit.ly/473NmIy
The Book of Life: Moving from a Sociology of Variables to a Sociology of Events – Princeton Laboratory for Artificial Intelligence Research Blog
bit.ly
February 20, 2026 at 8:25 PM
New Deep Dive from @anilananth.bsky.social on our blog:
@msalganik.bsky.social & colleagues developed a way to create and analyze an individual’s "Book of Life." The eventual goal? Train language models on data from millions of Books of Life to predict an individual's life outcome: bit.ly/473NmIy
@msalganik.bsky.social & colleagues developed a way to create and analyze an individual’s "Book of Life." The eventual goal? Train language models on data from millions of Books of Life to predict an individual's life outcome: bit.ly/473NmIy
Reposted by Princeton Laboratory for Artificial Intelligence
Congratulations to Sayash Kapoor on being awarded the Jacobus Fellowship, the highest honor bestowed on a graduate student at @princeton.edu. Kapoor will be honored in a ceremony this weekend.
engineering.princeton.edu/news/2026/02...
@sayash.bsky.social
engineering.princeton.edu/news/2026/02...
@sayash.bsky.social
February 18, 2026 at 4:57 PM
Congratulations to Sayash Kapoor on being awarded the Jacobus Fellowship, the highest honor bestowed on a graduate student at @princeton.edu. Kapoor will be honored in a ceremony this weekend.
engineering.princeton.edu/news/2026/02...
@sayash.bsky.social
engineering.princeton.edu/news/2026/02...
@sayash.bsky.social
Autonomous spacecraft are still a far off ideal 🚀
But Ryne Beeson and @stella.to are taking the first steps in that direction by finding the optimal trajectories to a given planet or moon with the help of machine learning: ai.princeton.edu/news/2025/ai...
But Ryne Beeson and @stella.to are taking the first steps in that direction by finding the optimal trajectories to a given planet or moon with the help of machine learning: ai.princeton.edu/news/2025/ai...
AI helps Princeton scientists plot the best paths for space exploration
When a spacecraft or probe is sent to explore Mars or do flybys of one of Saturn’s moons, it stays in constant contact with mission control back on Earth, where scientists recalculate and adjust as ne...
ai.princeton.edu
December 18, 2025 at 2:31 PM
Autonomous spacecraft are still a far off ideal 🚀
But Ryne Beeson and @stella.to are taking the first steps in that direction by finding the optimal trajectories to a given planet or moon with the help of machine learning: ai.princeton.edu/news/2025/ai...
But Ryne Beeson and @stella.to are taking the first steps in that direction by finding the optimal trajectories to a given planet or moon with the help of machine learning: ai.princeton.edu/news/2025/ai...
Here's a round up of the research being showcased by Princeton students, post-docs, research software engineers and faculty for the @neuripsconf.bsky.social this week: blog.ai.princeton.edu/2025/12/02/p...
December 2, 2025 at 6:01 PM
Here's a round up of the research being showcased by Princeton students, post-docs, research software engineers and faculty for the @neuripsconf.bsky.social this week: blog.ai.princeton.edu/2025/12/02/p...
Could geometry could be the key to unlock human-like intelligence in AI? 📐🧠
@timbuschman.bsky.social, @gautamreddy.bsky.social and Adel Ardalan dive into how neural networks use geometry to help models understand the world instead of just pattern-match it. On our blog now: bit.ly/49COIw8
@timbuschman.bsky.social, @gautamreddy.bsky.social and Adel Ardalan dive into how neural networks use geometry to help models understand the world instead of just pattern-match it. On our blog now: bit.ly/49COIw8
November 26, 2025 at 4:47 PM
Could geometry could be the key to unlock human-like intelligence in AI? 📐🧠
@timbuschman.bsky.social, @gautamreddy.bsky.social and Adel Ardalan dive into how neural networks use geometry to help models understand the world instead of just pattern-match it. On our blog now: bit.ly/49COIw8
@timbuschman.bsky.social, @gautamreddy.bsky.social and Adel Ardalan dive into how neural networks use geometry to help models understand the world instead of just pattern-match it. On our blog now: bit.ly/49COIw8
👋Meet the Postdoc!
Meet Andrew Nam, a researcher with our Natural and Artificial Minds initiative. His interests are abstract and rule-based reasoning, rapid learning, and out-of-distribution generalization in both biological and artificial intelligence systems.
ai.princeton.edu/news/2025/me...
Meet Andrew Nam, a researcher with our Natural and Artificial Minds initiative. His interests are abstract and rule-based reasoning, rapid learning, and out-of-distribution generalization in both biological and artificial intelligence systems.
ai.princeton.edu/news/2025/me...
Meet the Postdoc: Q&A with Andrew Nam
Meet The Postdoc is a monthly series highlighting the AI Lab’s postdoctoral researchers.Andrew Nam is a postdoctoral researcher in the Natural and Artificial Minds research initiative. His research in...
ai.princeton.edu
November 13, 2025 at 3:02 PM
👋Meet the Postdoc!
Meet Andrew Nam, a researcher with our Natural and Artificial Minds initiative. His interests are abstract and rule-based reasoning, rapid learning, and out-of-distribution generalization in both biological and artificial intelligence systems.
ai.princeton.edu/news/2025/me...
Meet Andrew Nam, a researcher with our Natural and Artificial Minds initiative. His interests are abstract and rule-based reasoning, rapid learning, and out-of-distribution generalization in both biological and artificial intelligence systems.
ai.princeton.edu/news/2025/me...
Reposted by Princeton Laboratory for Artificial Intelligence
Join us 11/6 with the Princeton AI Lab (@princetonainews.bsky.social) at 2-4 pm for the Policy Fellows Seminar. Learn from the researchers of the AI Lab who work closely with governmental agencies at the state & federal level to advance the responsible development and utilization of AI. Info 👇
November 5, 2025 at 5:16 PM
Join us 11/6 with the Princeton AI Lab (@princetonainews.bsky.social) at 2-4 pm for the Policy Fellows Seminar. Learn from the researchers of the AI Lab who work closely with governmental agencies at the state & federal level to advance the responsible development and utilization of AI. Info 👇
AI meets ancient art 🤝
In our latest blog post, read about the Princeton researchers combining machine learning and the principles of kirigami to transform flat surfaces into mesmerizing structures: blog.ai.princeton.edu/2025/10/31/u...
In our latest blog post, read about the Princeton researchers combining machine learning and the principles of kirigami to transform flat surfaces into mesmerizing structures: blog.ai.princeton.edu/2025/10/31/u...
October 31, 2025 at 6:10 PM
AI meets ancient art 🤝
In our latest blog post, read about the Princeton researchers combining machine learning and the principles of kirigami to transform flat surfaces into mesmerizing structures: blog.ai.princeton.edu/2025/10/31/u...
In our latest blog post, read about the Princeton researchers combining machine learning and the principles of kirigami to transform flat surfaces into mesmerizing structures: blog.ai.princeton.edu/2025/10/31/u...
Reposted by Princeton Laboratory for Artificial Intelligence
Excited to have this out and featured!
A new paper from @princetonneuro.bsky.social scientists shows people will forgo short-term rewards in favor of investing more effort when the opportunity for learning arises. The results could help researchers build better AI.
Read now: ai.princeton.edu/news/2025/hu...
Read now: ai.princeton.edu/news/2025/hu...
Human brains put in more effort when the opportunity to learn arises, the finding could help researchers build better AI
In a research paper published in PNAS, Princeton Neuroscience Institute researchers present their results finding that people will forgo short-term rewards in favor of investing more effort when the o...
ai.princeton.edu
October 30, 2025 at 2:03 AM
Excited to have this out and featured!
A new paper from @princetonneuro.bsky.social scientists shows people will forgo short-term rewards in favor of investing more effort when the opportunity for learning arises. The results could help researchers build better AI.
Read now: ai.princeton.edu/news/2025/hu...
Read now: ai.princeton.edu/news/2025/hu...
Human brains put in more effort when the opportunity to learn arises, the finding could help researchers build better AI
In a research paper published in PNAS, Princeton Neuroscience Institute researchers present their results finding that people will forgo short-term rewards in favor of investing more effort when the o...
ai.princeton.edu
October 28, 2025 at 4:40 PM
A new paper from @princetonneuro.bsky.social scientists shows people will forgo short-term rewards in favor of investing more effort when the opportunity for learning arises. The results could help researchers build better AI.
Read now: ai.princeton.edu/news/2025/hu...
Read now: ai.princeton.edu/news/2025/hu...
We welcomed 2 new associate research scholars this fall 🥳
Liam Fowl and @happybuzaaba1.bsky.social will be supporting the ongoing research at Princeton Language and Intelligence.
Read more: ai.princeton.edu/news/2025/ai...
Liam Fowl and @happybuzaaba1.bsky.social will be supporting the ongoing research at Princeton Language and Intelligence.
Read more: ai.princeton.edu/news/2025/ai...
AI Lab Welcomes Associate Research Scholars
This fall, two new associate research scholars join Princeton Language and Intelligence, one of the research initiatives under the AI Lab. The scholars will be supporting the ongoing research at PLI t...
ai.princeton.edu
October 20, 2025 at 7:56 PM
We welcomed 2 new associate research scholars this fall 🥳
Liam Fowl and @happybuzaaba1.bsky.social will be supporting the ongoing research at Princeton Language and Intelligence.
Read more: ai.princeton.edu/news/2025/ai...
Liam Fowl and @happybuzaaba1.bsky.social will be supporting the ongoing research at Princeton Language and Intelligence.
Read more: ai.princeton.edu/news/2025/ai...
🚨 Call for Seed Grant Proposals 🚨
The AI Lab is now reviewing seed grants to support research in large AI models, natural and artificial minds, AI for invention and other AI-related projects.
Deadline for proposals is 10/31. For more information: ai.princeton.edu/ai-lab/seed-...
The AI Lab is now reviewing seed grants to support research in large AI models, natural and artificial minds, AI for invention and other AI-related projects.
Deadline for proposals is 10/31. For more information: ai.princeton.edu/ai-lab/seed-...
October 14, 2025 at 6:31 PM
🚨 Call for Seed Grant Proposals 🚨
The AI Lab is now reviewing seed grants to support research in large AI models, natural and artificial minds, AI for invention and other AI-related projects.
Deadline for proposals is 10/31. For more information: ai.princeton.edu/ai-lab/seed-...
The AI Lab is now reviewing seed grants to support research in large AI models, natural and artificial minds, AI for invention and other AI-related projects.
Deadline for proposals is 10/31. For more information: ai.princeton.edu/ai-lab/seed-...
Reposted by Princeton Laboratory for Artificial Intelligence
📣 New preprint! We know humans are biased against AI-creativity. But what about LLMs, now often judging creativity in various contexts? Do they replicate, transform, or amplify this bias? We tested it. Turns out: AI is 2.5X more biased against its own work than humans. arxiv.org/pdf/2510.08831 🧵
arxiv.org
October 13, 2025 at 1:56 PM
📣 New preprint! We know humans are biased against AI-creativity. But what about LLMs, now often judging creativity in various contexts? Do they replicate, transform, or amplify this bias? We tested it. Turns out: AI is 2.5X more biased against its own work than humans. arxiv.org/pdf/2510.08831 🧵
Reposted by Princeton Laboratory for Artificial Intelligence
How can AI and other computational approaches help us to understand the distant past? @jrladd.com will share work from two cultural analytics projects that address these challenges by combining analog research methods and machine learning to study the print culture of the 16th and 17th centuries.
Sep 29: "Human and Machine Intelligence in Networks of Early Modern Print" with John Ladd (@jrladd.com), Assistant Professor in Computing and Information Studies at Washington & Jefferson College. Info/RSVP:
Human and Machine Intelligence in Networks of Early Modern Print
Human and Machine Intelligence in Networks of Early Modern Print
cdh.princeton.edu
September 27, 2025 at 2:08 PM
How can AI and other computational approaches help us to understand the distant past? @jrladd.com will share work from two cultural analytics projects that address these challenges by combining analog research methods and machine learning to study the print culture of the 16th and 17th centuries.
WATCH: Princeton assistant prof of computer science @tridao.bsky.social on the Unsupervised Learning podcast
www.youtube.com/watch?v=xlSa...
www.youtube.com/watch?v=xlSa...
Tri Dao: The End of Nvidia's Dominance, Why Inference Costs Fell & The Next 10X in Speed
YouTube video by Unsupervised Learning: Redpoint's AI Podcast
www.youtube.com
September 25, 2025 at 7:40 PM
WATCH: Princeton assistant prof of computer science @tridao.bsky.social on the Unsupervised Learning podcast
www.youtube.com/watch?v=xlSa...
www.youtube.com/watch?v=xlSa...
This fall, we're welcoming 8 new postdocs!
From reinforcement learning to human-AI collaboration, their work will power forward our initiatives.
Meet them and learn more about their research: ai.princeton.edu/news/2025/ai...
From reinforcement learning to human-AI collaboration, their work will power forward our initiatives.
Meet them and learn more about their research: ai.princeton.edu/news/2025/ai...
AI Lab Welcomes New Postdocs
The Princeton Laboratory for Artificial Intelligence (AI Lab) is welcoming eight new postdoctoral researchers this fall. The incoming postdocs will contribute to the research coming from the AI Lab’s ...
ai.princeton.edu
September 23, 2025 at 2:22 PM
This fall, we're welcoming 8 new postdocs!
From reinforcement learning to human-AI collaboration, their work will power forward our initiatives.
Meet them and learn more about their research: ai.princeton.edu/news/2025/ai...
From reinforcement learning to human-AI collaboration, their work will power forward our initiatives.
Meet them and learn more about their research: ai.princeton.edu/news/2025/ai...