BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260605T154542Z
LOCATION:Bldg. 8 - Room B 102
DTSTART;TZID=Europe/Stockholm:20260630T141500
DTEND;TZID=Europe/Stockholm:20260630T144500
UID:submissions.pasc-conference.org_PASC26_sess148_msa253@linklings.com
SUMMARY:Responsible and Ethical AI in HPC: The Case for Epistemic Infrastr
 ucture
DESCRIPTION:Monica Morrison (NSF NCAR)\n\nThe HPC community is rapidly adv
 ancing AI capabilities, but technical maturity alone is not sufficient for
  its use in Earth system science. Realizing AI's genuine potential to evol
 ve scientific methods and knowledge requires building epistemic infrastruc
 ture in parallel: the frameworks and practices that allow us to evaluate d
 ata, tool, and model reliability, assess fitness-for-purpose, and quantify
  uncertainty before deploying AI in scientific workflows.\n\nThis talk pre
 sents three interlinked components of such an infrastructure, developed in
  the context of Earth system science: data fitness-for-purpose frameworks;
  Model Readiness level to evaluate AI/ML tool across reliability, interpre
 tability, and purpose-fitness dimensions; and uncertainty quantification p
 ractices to track sources of epistemic risk through computational pipeline
 s. Together, they shift the question we use to address AI in science to "h
 ow do we know this AI-enhanced outputs are adequate for their intended use
 —especially when results, such as weather model outputs, might inform cons
 equential decisions?" Without parallel investment in the development and i
 mplementation of epistemic infrastructure, the drive toward efficiency ris
 ks greatly outpacing our capacity for accountability, reproducible science
 , and transparency, rendering AI not a panacea for scientific advancement,
  but a new source of epistemic vulnerability.\n\nDomain: Computational Met
 hods and Applied Mathematics\n\nSession Chairs: Nick Brown (EPCC) and Rui 
 Apostolo (EPCC)\n\n
END:VEVENT
END:VCALENDAR