Replika in the Metaverse: the moral problem with empathy in ‘It from Bit’
Research output: Contribution to journal › Article › peer-review
Standard Standard
In: AI and Ethics, Vol. 3, No. 4, 11.2023, p. 1433-1445.
Research output: Contribution to journal › Article › peer-review
HarvardHarvard
APA
CBE
MLA
VancouverVancouver
Author
RIS
TY - JOUR
T1 - Replika in the Metaverse: the moral problem with empathy in ‘It from Bit’
AU - McStay, Andrew
PY - 2023/11
Y1 - 2023/11
N2 - This paper assesses claims of computational empathy in relation to existing social open-ended chatbots and intention that these chatbots will feature in emergent mixed reality contexts, recently given prominence due to interest in the Metaverse. Against the background of increasing loneliness within society and use of chatbots as a potential remedy for this, the paper considers two leading current social chatbots, Replika and Microsoft's Xiaoice, their technical underpinnings, empathetic claims and properties that have scope to scale into the Metaverse (if it coheres). Finding scope for human benefit from social chatbots, the paper highlights problematic reliance on self-disclosure to sustain the existence of chatbots. The paper progresses to situate Microsoft's empathetic computing framework in relation to philosophical ideas that inform Metaverse speculation and construction, including Wheeler's 'It from Bit' thesis that all aspects of existence may be computed, Chalmers' philosophical championing that virtual realities are genuine realities, Bostrom's proposal and provocation that we might already be living in a simulation, and longtermist belief that future complex simulations need to be protected from decisions made today. Given claims for current and nascent social chatbots, belief in bit-based possible and projected futures, and industrial buy-in to these philosophies, this paper answers whether computational empathy is real or not. The paper finds when diverse accounts of empathy are accounted for, whilst something is irrevocably lost in an 'It from Bit' account of empathy, the missing components are not accuracy or even human commonality of experience, but the moral dimension of empathy.
AB - This paper assesses claims of computational empathy in relation to existing social open-ended chatbots and intention that these chatbots will feature in emergent mixed reality contexts, recently given prominence due to interest in the Metaverse. Against the background of increasing loneliness within society and use of chatbots as a potential remedy for this, the paper considers two leading current social chatbots, Replika and Microsoft's Xiaoice, their technical underpinnings, empathetic claims and properties that have scope to scale into the Metaverse (if it coheres). Finding scope for human benefit from social chatbots, the paper highlights problematic reliance on self-disclosure to sustain the existence of chatbots. The paper progresses to situate Microsoft's empathetic computing framework in relation to philosophical ideas that inform Metaverse speculation and construction, including Wheeler's 'It from Bit' thesis that all aspects of existence may be computed, Chalmers' philosophical championing that virtual realities are genuine realities, Bostrom's proposal and provocation that we might already be living in a simulation, and longtermist belief that future complex simulations need to be protected from decisions made today. Given claims for current and nascent social chatbots, belief in bit-based possible and projected futures, and industrial buy-in to these philosophies, this paper answers whether computational empathy is real or not. The paper finds when diverse accounts of empathy are accounted for, whilst something is irrevocably lost in an 'It from Bit' account of empathy, the missing components are not accuracy or even human commonality of experience, but the moral dimension of empathy.
KW - Augmented reality
KW - Chatbot
KW - Empathy
KW - Metaverse
KW - Mixed reality
KW - Replika
KW - Xiaoice
U2 - 10.1007/s43681-022-00252-7
DO - 10.1007/s43681-022-00252-7
M3 - Article
C2 - 36573214
VL - 3
SP - 1433
EP - 1445
JO - AI and Ethics
JF - AI and Ethics
SN - 2730-5961
IS - 4
ER -