BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.com//python-asia-2026//speaker//DDF7LU
BEGIN:VTIMEZONE
TZID:PST
BEGIN:STANDARD
DTSTART:20000101T000000
RRULE:FREQ=YEARLY;BYMONTH=1
TZNAME:PST
TZOFFSETFROM:+0800
TZOFFSETTO:+0800
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-python-asia-2026-L3RNT9@pretalx.com
DTSTART;TZID=PST:20260321T100500
DTEND;TZID=PST:20260321T105000
DESCRIPTION:As Large Language Models (LLMs) expand into global markets\, th
 ey often hit a "Nuance Gap" or a failure to distinguish literal meaning fr
 om cultural context. This presentation examines the technical hurdles of b
 uilding culturally competent AI through two lenses\, namely\, linguistic a
 mbiguity (sarcasm) and localized safety (toxicity). Using the Philippines 
 as a case study\, we identify four critical hurdles: (1) the Linguistic In
 version Problem\, where sarcasm flips intended sentiment\; (2) the Context
  Vacuum\, where text lacks the "cultural scaffolding" necessary for interp
 retation\; (3) the Data Desert of low-resource languages\; and (4) the Wes
 tern-Centricity of standard safety filters. We propose a roadmap for resea
 rchers to move beyond literal translation toward AI that respects the "uns
 poken" and "unseen" nuances of regional identities.
DTSTAMP:20260501T070413Z
LOCATION:Teresa Yuchengco Auditorium (Main Hall)
SUMMARY:[Keynote] Architectures of Ambiguity: Mapping the Technical Hurdles
  of Cultural Sensitivity in Localized LLMs - Charibeth Cheng
URL:https://pretalx.com/python-asia-2026/talk/L3RNT9/
END:VEVENT
END:VCALENDAR