Aisvarya Chandrasekar

I’m a computational journalist and fellow at the Tow Center for Digital Journalism, where I study how AI is changing the way people find and understand news, and how large language models can support reporting.

I’ve tested whether AI browsers get past paywalls, mapped industry influence in scientific research, built tools that help newsroom researchers find polls, and analyzed how long it takes slang to reach the NYT crossword.

I love making things and trying out new mediums and methods. I’m usually picking up a new hobby or working on a side quest.

AI & the news

As more people get their news through AI platforms, the choices AI companies make, their relationship with publishers, and the limits of their technology decide which reporting gets seen and credited. I design experiments and write about how AI is reshaping journalism. Most of this work is done in collaboration with Klaudia Jaźwińska.

More AI & news work
  • Illustration accompanying reporting on chatbot memory and personalization

    Your Chatbot’s Memory of You Can Shape the Information You See

    AI companies promise that remembering users makes chatbots more helpful, but how do those memories shape the information people receive? I reported on research into how memory and personalization can reinforce users’ beliefs, shape answers, and leave people with less control over stored information than companies suggest.

  • Illustration accompanying reporting on AI web search

    AI Chatbots Can Search the Web—So Why Don’t They?

    AI chatbots can search for current information, but they don’t always choose to. We explain why these tools sometimes skip web search, how that can lead to confidently wrong answers about the news, and why those failures demand accountability from platforms.

  • ChatGPT Atlas logo displayed on a phone

    How AI Browsers Sneak Past Blockers and Paywalls

    We tested how AI browsers access news articles that publishers try to restrict. Our reporting explains how agents can appear as ordinary browser sessions and read text hidden by some paywalls, complicating publishers’ efforts to control how their work is used.

Other projects

Illustration for The Examination’s investigation into Red Bull’s ties to energy drink research

July 2026

How AI helped us uncover Red Bull’s influence on energy drink research

We used an LLM to map financial ties between energy drink companies and researchers for The Examination. We verified the extracted disclosures and built a network graph to help reporters find leads.

Balancing the books – equitable library funding in Ohio

July 2023

Balancing the Books

Ohio’s library funding system was designed to distribute resources more equitably across counties. I examined how that model changed as libraries increasingly turned to local levies, and what the shift meant for communities with different property-tax bases.

Slang in the NYT crossword

July 2022

From PHAT to DEADASS

Does the New York Times crossword’s idea of modern language actually keep up with the way people speak? I collected crossword clues from XWordInfo to find out.

Map highlighting the Hudson Greenway bike route along Manhattan

August 2022

Fear and Luminescence on a New York Bike Trail

My morning bike ride along the Hudson Waterfront Greenway felt very different after dark. I documented one night-time commute to explore how lighting changed my experience of the route.

Speaking & presentations