Megha Srivastava

102 posts

Megha Srivastava

@megha_byte

👩🏽‍💻🏔📷🌲🎨 phd student in computer science @stanfordailab + applied cryptography group 🧩

Los Angeles, CA Katılım Kasım 2010

612 Takip Edilen645 Takipçiler

Sabitlenmiş Tweet

Megha Srivastava@megha_byte·25 Ağu

Ever wanted to generate smokey logograms from the heptapod language in the awesome sci-fi movie "Arrival" (2016)? Probably not 😅 But a few people asked for the text-to-image model I trained, so it's now on @replicate ! replicate.com/meghabyte/arri…

English

315

104.3K

Megha Srivastava retweetledi

Steph Chambers@StephChambers76·20h

March cherry blossom Madness. Tacoma, Washington

English

304

3.4K

48.9K

Megha Srivastava@megha_byte·15 Oca

Check out our paper taking a broader look at emergent misalignment in large language models, published today in @Nature !

Owain Evans@OwainEvans_UK

Today we publish an extended paper on emergent misalignment in Nature. We communicate our original insecure-code results to a broad audience and incorporate the findings from many other researchers. It's open access! nature.com/articles/s4158…

English

2.3K

Megha Srivastava retweetledi

ProPublica@propublica·9 Kas

Blue Cross had refused to pay thousands of claims for the center’s patients, but on several occasions execs at the insurance company had signed special one-time deals with the center to pay for their wives’ cancer treatment. propublica.org/article/blue-c…

English

710

986

53.1K

Megha Srivastava@megha_byte·12 Kas

@shershman @claudeai @james_y_zou Unfortunately I don't think the presentation was shared, but my slides are here: meghabyte.github.io/echo_spotlight… as well as a blog post about the research process! meghabytes.substack.com/p/notes-from-a…

English

Steve Hershman@shershman·22 Eki

@megha_byte @claudeai @james_y_zou Your talk sounded really interesting - will it be recorded?

English

Megha Srivastava@megha_byte·22 Eki

I'll be discussing work on using AI for pharmacovigilance (led by @claudeai 😉 ) as a Spotlight in #Agents4Science today at 11:00am PST! It was really fun participating in @james_y_zou's experiment on what happens if we let an LM lead a research project agents4science.stanford.edu

English

10.3K

Megha Srivastava@megha_byte·11 Eki

@alexisjross @lucy3_li @jacobandreas @DoctorJei Likewise!! 🤩

English

Alexis Ross@alexisjross·10 Eki

@lucy3_li @megha_byte @jacobandreas @DoctorJei 🥹🥹🥹 very lucky I got to work with @megha_byte 🫶🏻

English

175

Megha Srivastava@megha_byte·7 Eki

New preprint on AI + Education! 🍎 “Modeling Student Learning with 3.8M Program Traces” 💻 When students code, their edits tell a story about their reasoning process: exploring, debugging, and tinkering 🧠 What can LMs learn from training on student edit sequences? 📚

English

211

55.6K

Megha Srivastava@megha_byte·9 Eki

@shayekhbinislam @alexisjross @jacobandreas @DoctorJei Most of the data is from pre-2022 (when chatgpt was released), and from an isolated env on pencilcode.net — so unlikely any GenAI tools were used! But it would be cool to study their effects on learning 🙂

English

145

Shayekh Islam@shayekhbinislam·8 Eki

@megha_byte @alexisjross @jacobandreas @DoctorJei Does these students take help of GenAI tools while coding or the study is performed in an isolated environment?

English

243

Megha Srivastava retweetledi

Joseph Fasano@Joseph_Fasano_·8 Eki

I can't stop thinking about Adrienne Rich's poem about AI, written in 1961.

English

287

1.6K

61K

Megha Srivastava@megha_byte·7 Eki

We also release code/data at github.com/meghabyte/penc… and are grateful to the PencilCode team for their support!

English

702

Megha Srivastava@megha_byte·7 Eki

💡 Takeaway: Many properties of code are properties of the student coder. LMs trained on students’ edit traces are more generalizable, controllable, and predictive of diverse student behavior. Paper: arxiv.org/abs/2510.05056 Joint work w/ @alexisjross , @DoctorJei, @jacobandreas

English

816

Megha Srivastava retweetledi

Merriam-Webster@MerriamWebster·26 Eyl

We are thrilled to announce that our NEW Large Language Model will be released on 11.18.25.

English

924

30.6K

204.1K

21.4M

Megha Srivastava@megha_byte·3 Eyl

highly recommend @zednelson ‘s The Anthropocene Illusion book and photo series! npr.org/sections/goats…

English

452

Megha Srivastava retweetledi

Alvaro Bedoya@BedoyaUSA·30 Ağu

If you read this story - about how Medicare Advantage plans are about to require prior approvals, and likely use AI to deny those prior approvals, you should learn a name: Gary Bent

Sen. Bernie Sanders@SenSanders

No, President Trump. Forcing seniors with traditional Medicare to get prior approval from AI companies before they can receive the health care they need won’t make America healthy again. It will make AI companies even richer by denying care to seniors. DON’T CUT MEDICARE.

English

368

991

88.9K

Megha Srivastava retweetledi

Deb Raji@rajiinio·29 Ağu

... the commercial incentives and capital requirements required to build AGI make it impossible to safely build "AGI"? 😶

English

116

4.6K

Megha Srivastava retweetledi

Ben Recht@beenwrekt·28 Ağu

I wrote about the tragic death of Adam Raine and the venal negligence of "AI Safety." argmin.net/p/the-banal-ev…

English

28.9K

Keşfet

@Nature @shershman @claudeai @james_y_zou @alexisjross @lucy3_li @jacobandreas @DoctorJei