Eshwar Chandrasekharan

PeerPen, a writing assistance tool embedded within a Reddit-like interface, supporting two main features: draft generation and revision of user-written responses, is built and it is found that PeerPen reduced the burden of composing responses and increased confidence in offering support.

Mislabeling AI-generated financial advice can paradoxically boost its perceived quality, highlighting the power of source attribution in shaping trust.

Open-source LLMs, trained with Reddit community feedback, can match the performance of proprietary models in generating supportive mental health responses, offering a privacy-preserving alternative.

LLM-based query rewriting in RAG can reduce retrieval bias by over 50%, but breaks down when biases combine adversarially, revealing the limits of query-side interventions.

Simulations that look realistic aren't enough: to reliably test governance interventions, we need to move to causal simulations that can support policy changes.

AI-agent communities aren't just pale imitations of human ones; they're structurally and linguistically distinct, exhibiting extreme inequality and homogenization driven by identifiable agent-level stylistic outliers.

Generative search engines create "answer bubbles" by selectively citing and framing information, leading to divergent information realities compared to traditional search.