BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:IC Colloquium: General Architectures for Specialized AI Interactio
 ns
DTSTART:20260119T101500
DTEND:20260119T111500
DTSTAMP:20260916T063751Z
UID:6af395eb0c73444f403cbb7f8c9c0f8ca114e694445d765c23a30950
CATEGORIES:Conferences - Seminars
DESCRIPTION:By: Michelle Lam - Stanford University\nIC Faculty candidate\n
 \nAbstract\nUsing a large language model is like talking to a personal ass
 istant\, confidant\, topic expert\, and copyeditor all at once: and that
 ’s a problem. When an LLM is one big mixed metaphor\, people must manage
  incompatible mental models about how it works\, hindering their ability t
 o interpret and control model behavior. In this talk\, I argue that redesi
 gning AI abstractions for end-user control can enable users to redirect ge
 neric AI systems to realize their specific\, distinctive goals. I demonstr
 ate this concept by introducing just-in-time (JIT) objectives\, which spec
 ialize outputs to a particular user by inducing user objectives from obser
 ved interaction traces. JIT objectives enable on-the-fly generation of sof
 tware tools that visualize a research statement’s logical argument\, tes
 t alternate color palettes for a figure\, or provide feedback based on rel
 evant academic experts. I then demonstrate how this approach also enables 
 interventions into other AI systems such as social media rankers (Societal
  Objective Functions) and concept induction from unstructured text data (L
 LooM). This work argues that user-controllable AI objectives are a viable 
 strategy to combat the issues of generic AI interactions.\n\nBio\nMichelle
  Lam is a Computer Science PhD candidate at Stanford University in the Hum
 an-Computer Interaction Group. She designs\, builds\, and deploys novel sy
 stems for user-controllable AI\, shifting the work of defining and evaluat
 ing AI objectives closer to the time and place where end users interact wi
 th AI. Michelle publishes at top HCI venues such as ACM CHI\, UIST\, and C
 SCW\, where she has received Best Paper Awards (CSCW ’24\, CHI ’22)\, 
 an Impact Recognition (CSCW ’24)\, and a Best Paper honorable mention (C
 SCW ’23). She was recognized as a Rising Star in EECS\, Stanford Interdi
 sciplinary Graduate Fellow\, and Siebel Scholar.\n\nMore information
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420 https://epfl.zoom.us/
 j/63790535207
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR