BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Drupal: Date iCal//NONSGML kigkonsult.se iCalcreator 2.18//
METHOD:PUBLISH
X-WR-CALNAME;VALUE=TEXT:Calendar of Events
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:STANDARD
DTSTART:20261101T020000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
RDATE:20271107T020000
TZNAME:EST
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20270314T020000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:calendar.71863.field_date.0@calendar.lafayette.edu
DTSTAMP:20260919T122928Z
DESCRIPTION:Concerns about artificial intelligence—particularly systems bas
 ed on  \nmachine learning models whose internal operations are opaque to h
 uman  \nunderstanding—are frequently framed in terms of accountability. Bu
 t these  \nconcerns are often left underspecified\, encompassing a heterog
 eneous set of  \nissues that call for distinct responses.\n\nThis talk foc
 uses on a specific subset of accountability concerns: the  \n“who” questio
 ns. We distinguish between two questions: When an AI system  \ncauses harm
 \, who is to blame\, and who bears the obligation to provide  \ncompensati
 on? We argue that the answers to these two questions require  \ndifferent 
 information and don’t entail one another.\n\nThen we investigate whether t
 he opacity of contemporary machine learning  \nmodels undermines our capac
 ity to hold the relevant actors accountable in  \nthese two senses of acco
 untability. What is required\, we argue\, is not  \nunderstanding of the i
 nternal processes of the particular machine learning  \nmodel deployed\, b
 ut rather\, a general understanding of how upstream decisions  \nin the de
 velopment process shape downstream model behavior and internal  \nprocesse
 s. We conclude with various practical implications of our discussion.  \nA
 mong others\, we argue for a pro tanto moral obligation to document  \ncon
 textual and historical information in the process of developing machine  
 \nlearning models.
DTSTART;TZID=America/New_York:20261111T161500
DTEND;TZID=America/New_York:20261111T173000
LAST-MODIFIED:20260909T150538Z
LOCATION:Pardee Hall Room 321
SUMMARY:The Ethics of AI: (How) Does Accountability Require Understanding M
 achine  \nLearning Models?
URL;TYPE=URI:https://calendar.lafayette.edu/node/71863
END:VEVENT
END:VCALENDAR