BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20251104T023446EST-670158FxHv@132.216.98.100
DTSTAMP:20251104T073446Z
DESCRIPTION:Title: Moral Learning and Artificial Intelligence.\n\nCafé/bisc
 uits dès 15h.Traditional approaches to moral development have emphasized t
 he explicit teaching of norms\, e.g.\, via parental instruction\, or the a
 cquisition of behavioral dispositions by “social learning”\, e.g.\, via in
 fant imitation and modeling of observed behaviors\, or progression through
  a fixed set of developmental “stages”. But what if we understood moral le
 arning as closer to causal learning and the development of commonsense phy
 sics? Developmental evidence suggests that infants early on begin to model
  their physical environment and its possibilities (Gopnik & Schulz\, 2004)
 \, using observation but receiving very limited explicit instruction or ex
 ternal reinforcement. Similarly\, there is evidence that infants early on 
 begin learning a kind of commonsense psychology that enables them to model
  others’ behavior in terms of intentional states\, once again\, using obse
 rvation but very limited explicit instruction or external reinforcement (W
 ellman\, 2014). These internal models enable infants to interact reasonabl
 y successfully with their physical and social environment even if they are
  unable to articulate the causal or psychological principles involved—the 
 knowledge underlying these capacities is therefore generalizable despite b
 eing implicit\, and so is spoken of as intuitive. Internal models are not 
 limited to causal and predictive information\, however\, but also appear t
 o encode evaluative information\, including evaluation of possible actions
  or third-party social interactions for such features as helpfulness\, har
 m\, knowledgeability\, and trustworthiness (Hamlin et al.\, 2011\; Doebel 
 & Koenig\, 2013). When combined with an implicit capacity to empathically 
 simulate the mental states of others\, these evaluative capacities can und
 erwrite a kind of intuitive learning of commonsense morality. Such learnin
 g occurs without much explicit instruction in moral principles\, yet with 
 a capacity to generalize and with some degree of moral autonomy—so that by
  age 3-4\, infants will resist conforming to imposed rules that involve ha
 rm or unfairness toward others (Turiel\, 2002). To be genuinely intelligen
 t\, artificial systems will need to possess the kinds of intuitive knowled
 ge involved in commonsense physics and psychology. And to be both autonomo
 us and trustworthy\, artificial systems will need to be able to evaluate s
 ituations\, actions\, and agents in the terms of such categories of common
 sense morality as helpfulness\, harm\, knowledgeability\, and trustworthin
 ess. Deep learning approaches suggest how intuitive knowledge of the kind 
 involved in predictive learning might be acquired and represented\, withou
 t being “programmed in” or explicitly taught. How might further developmen
 ts of these approaches make possible the acquisition of intuitive evaluati
 ve knowledge of the kind involved in commonsense epistemic or moral assess
 ment?\n\n\n\n
DTSTART:20180118T203000Z
DTEND:20180118T203000Z
LOCATION:Room Z330\, CA\, UdeM\, Pavillon Claire McNicoll
SUMMARY:Peter Railton (University of Michigan)
URL:/mathstat/channels/event/peter-railton-university-
 michigan-283824
END:VEVENT
END:VCALENDAR