Mr. @SecMullinDHS, as you step into your new position I ask you to provide particular consideration to the status of Iranian students and professionals in the United States who may be at risk of immigration complications due to Presidential Proclamations 10949 and 10998. Many of these individuals fled the Islamic Republic’s because they believe in freedom, democracy, and the values this country stands for. They are also contributing to the United States every day.
They arrived in this country legally, followed the rules, respect this country and its system of law and order and are among the most dedicated and accomplished contributors to American society. I ask that you and your department consider these factors, plus the personal and political risk many of these young people would face if sent back to the criminal regime in Iran.
America’s concerns for its national security and the threat posed by the Islamic regime to the homeland are entirely justified. To that end, my office stands ready to provide you and your colleagues with regime operatives, propaganda agents, and influence networks for investigation and removal from the United States.
Thank you for your consideration.
.@POTUS: Today, I send my best wishes to every American celebrating Nowruz.
Nowruz marks the start of the Persian New Year, celebrates the arrival of spring, and acknowledges the eternal triumph of light over darkness.
https://t.co/CCweyIqG7S
Secretary @marcorubio
Iranian immigrants have long contributed to the U.S. economy, especially in STEM, medicine and AI. Most strongly oppose the Islamic regime and support the goal of a strong and secure USA under President Trump and your leadership.
Over the past few months, many have faced serious hardship due to the USCIS hold. They are here legally and often hold advanced degrees. Returning them to Iran would place them in danger and risk sending highly trained experts in strategic fields back to the Iranian regime.
We would greatly appreciate it if you could revisit this decision.
Thank you for your consideration.
The Iranian regime does not reflect the people of Iran, nor their culture rooted within a deep history. I know of no other country where there’s a bigger difference between the people who lead the country and the people who live there.
How does noise from LLM-generated annotations effect AI classification performance? A new simulation study reveals systematic, prevalence-dependent biases in model evaluation. https://t.co/5rL4ZbNGL0 @ChavoshiSmr#LargeLanguageModels#AI#ML
🗓️ Exactly one week ago to the hour, #Iran fell into digital darkness as authorities imposed a national internet blackout.
Through the following days Iranians continued to protest and demand liberty despite a draconian crackdown.
At 168 hours, data show the shutdown is ongoing.
⚠️ Update: Metrics show #Iran remains offline as the country wakes to another day of digital darkness.
With the internet blackout now past its 132nd hour, early reports indicate thousands of casualties. The true extent of the killings is obscured by the absence of connectivity.
What’s unfolding in Iran may be the most consequential and underreported story in the world. Free elections there won’t just be a human rights victory. The fall of the Ayatollahs will release tens of millions of people, who come from one of the most creative civilizations in history. Iranian culture has shaped art, science and innovation for centuries. Suppressing it hurts not just Iranians, but humanity as a whole.
⚠️ Update: #Iran has now been offline for 12 hours with national connectivity flatlining at ~1% of ordinary levels, after authorities imposed a national internet blackout in an attempt to suppress sweeping protests while covering up reports of regime brutality 📉
Impact of Label Noise from Large Language Models Generated Annotations on Evaluation of Diagnostic Model Performance https://t.co/5rL4ZbN8Vs @ChavoshiSmr#LargeLanguageModels#AI#ML
In low-prevalence, small drops in LLM specificity can make a near-perfect model look awful.
We built an interactive calculator where you can plug in prevalence + LLM sensitivity/specificity and see how much “model performance” is actually label noise:
https://t.co/y3RYCv6Mo8
I’ve been using LLMs to extract labels from radiology reports for a while. This paper came from realizing how easily that can bias evaluation.
The key lesson: labeling with LLMs must be prevalence-aware!
LLM-generated labels can introduce disease prevalence-dependent systemic bias into AI binary classification model performance evaluation https://t.co/5rL4ZbN8Vs @ChavoshiSmr#LargeLanguageModels#ML#MachineLearning
Our new paper challenges the "one size fits all" myth of AI in medicine. We show that foundation models struggle with fine-grained diagnostic tasks like knee osteoarthritis, proving that specialized deep learning is still superior in domains with limited training data
#Radiology
🚨 Excited to share our new paper in @LancetDigitalH!
LLMs have revolutionized language generation, but what about image generation in medicine? 🏥🔬
1/
@YaleRadRes@MayoRadiology @EmoryRadiology
📅 Register by July 31, 2025 for the Healthcare AI Lab’s Summer School & Datathon at https://t.co/8trlrvi6oF! Don’t miss this chance to explore cutting-edge medical AI. #RadiologyAI#MachineLearning
🚀 Join us for the 3rd Medical AI Summer School, Symposium & Datathon (Aug 17-24, 2025)! Dive into Health AI with foundation models & multimodal data. Expect keynotes, hands-on training & more! #MedicalAI#HealthTech
https://t.co/j3QJPZSc2i