BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//aiweek.boston//Schedule//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:Behavior adaptation strategies for language models in high sensitivity and vulnerable-user contexts
X-WR-TIMEZONE:America/New_York
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20070311T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20071104T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
UID:behavior-adaptation-strategies-for-language-models-in-high-sensitivity-and-vulne@aiweek.boston
DTSTAMP:20260913T214814Z
SEQUENCE:0
STATUS:CONFIRMED
DTSTART;TZID=America/New_York:20260927T163000
DTEND;TZID=America/New_York:20260927T180000
SUMMARY:Behavior adaptation strategies for language models in high sensit
 ivity and vulnerable-user contexts
DESCRIPTION:This meetup is designed as a host-led\, structured panel conv
 ersation rather than a sequence of presentations. The hosts will introduc
 e each topic\, pose questions and scenarios\, and guide the panelists thr
 ough different perspectives\, trade-offs\, and practical experiences. The
  goal is an open technical discussion where ideas can be compared\, chall
 enged\, and developed through conversation.\n\nAs language models become 
 more capable and are increasingly deployed in high-sensitivity and vulner
 able-user contexts\, we will focus on a central question: How do we adapt
  these systems to different users and situations without weakening the sa
 feguards they depend on?\n\nThe hosts will guide the conversation through
  four discussion areas:\n\nBoundary Safeguards: Where should behavioral a
 nd safety boundaries be drawn? How should teams enforce and monitor them 
 when models operate in high-sensitivity and vulnerable-user contexts?\n\n
 Experimental Rigor: What does responsible experimentation look like? The 
 panel will discuss A/B testing\, controlled experiments\, and staged depl
 oyments as ways to evaluate adaptations while detecting unintended behavi
 ors and risks.\n\nAdaptation Playbooks: How should models be adapted for 
 different users\, workflows\, and contexts while preserving essential saf
 eguards? Panelists will compare approaches involving prompting\, model ad
 aptation\, and application-level controls—and discuss where each approach
  works or breaks down.\n\nMeasuring Success: What should teams measure be
 yond accuracy? The conversation will examine safety\, fairness\, robustne
 ss\, user outcomes\, and unintended behavior\, including how these measur
 es influence production decisions.\n\nThroughout the meetup\, the hosts w
 ill use these themes as a structure for the conversation rather than a fi
 xed agenda. Panelists will be encouraged to compare approaches\, surface 
 disagreements\, share practical lessons\, and explore the tensions betwee
 n adaptability and safeguards.
LOCATION:The Foundry \nCambridge\, Roger's st Cambridge ma 02141
URL:https://aiweek.boston/schedule/behavior-adaptation-strategies-for-lan
 guage-models-in-high-sensitivity-and-vulne
BEGIN:VALARM
ACTION:DISPLAY
DESCRIPTION:Behavior adaptation strategies for language models in high se
 nsitivity and vulnerable-user contexts
TRIGGER:-P1D
END:VALARM
BEGIN:VALARM
ACTION:DISPLAY
DESCRIPTION:Behavior adaptation strategies for language models in high se
 nsitivity and vulnerable-user contexts
TRIGGER:-PT1H
END:VALARM
END:VEVENT
END:VCALENDAR