http://localhost:4000/ Researcher, Engineer, Musician 2026-07-16T00:21:39-04:00 Lillian Tsai http://localhost:4000/ Jekyll © 2026 Lillian Tsai /assets/img/favicons/favicon.ico /assets/img/favicons/favicon-96x96.png 2026-06-12T00:00:00-04:00 2026-06-12T00:00:00-04:00 http://localhost:4000/posts/vimendelssohn/ Lillian Tsai

2026-06-01T01:00:00-04:00 2026-06-01T01:00:00-04:00 http://localhost:4000/posts/tpdp/ Lillian Tsai

We built ContinuousBench, a benchmark for whether DP synthetic text retains the corpus-specific knowledge that motivated using a sensitive dataset in the first place. Existing benchmarks like IMDB and OpenReview are saturated — train-on-real beats no-training by only a few points (51% → 62% on OpenReview, 94% → 97% on IMDB), leaving no room to distinguish DP synthesis methods. On ContinuousBen...

2026-05-11T01:00:00-04:00 2026-05-11T01:00:00-04:00 http://localhost:4000/posts/workflowstore/ Lillian Tsai

The dominant paradigm for AI agents is an “on-the-fly” loop in which agents synthesize plans and execute actions within seconds or minutes in response to user prompts. We argue that this paradigm short-circuits disciplined software engineering (SE) processes – iterative design, rigorous testing, adversarial evaluation, staged deployment, and more – that have delivered the (relatively) reliable ...

2026-04-21T01:00:00-04:00 2026-04-21T01:00:00-04:00 http://localhost:4000/posts/gaap/ Lillian Tsai

AI agents promise to serve as general-purpose personal assistants for their users, which requires them to have access to private user data (e.g., personal and financial information). This poses a serious risk to security and privacy. Adversaries may attack the AI model (e.g., via prompt injection) to exfiltrate user data. Furthermore, sharing private data with an AI agent requires users to trus...

2026-04-18T00:00:00-04:00 2026-04-18T00:00:00-04:00 http://localhost:4000/posts/mffkodalyshosti/ Lillian Tsai