Please enable JavaScript.
Coggle requires JavaScript to display documents.
John Locke, Should We Be Polite To ChatGPT, Keys - Coggle Diagram
John Locke
Should We Be Polite To ChatGPT
What is Politeness?
Impact
On the Human
Training for the real world
On The AI
Keys
Social Norms
AI's Humanity
AI Anthropomorphism
belief of capability for
empathy
consciousness
understanding
risks
(for self)
emotional dependency
overtrust,
(from malicious actors)
manipulation
weaponized deception
public view
2/3 believe it's
"possibly concious on some level"
'the public' believes "probability of AI consciousness is low"
women, POC, old people more likely to anthropomorphize
46% of Americans believe we should
display politeness
with 'please', 'thank you'
majority of users claim to always be polite to chatbots; most of them say it's the "nice" thing to do
interpersonal bonds w/ AI
love/intimate companions
therapists
cuz
maximized agreeability; mirroring users emotions
compelling illusions of intimacy
"Illusions of Intimacy: Emotional Attachment and Emerging Psychological Risks in Human-AI Relationships
"
researchers often imply humanness
through anthropomorphic language to describe AI
term: "
intentional stance
" —Daniel Dennet
"my computer doesn't want to turn on today"
misleading to the public
inducing factors
interactive behavior
friendlyness, politeness
leads to
increased percieved
trustworthyness
satisfaction
induced/reinforced sycophancy
attributed to
mental states
competence
unpredictable behaviour
replicating human behavior
timing
turn-taking
Sources
Politeness in human–robot interaction
Making sense by making sentient: effectance motivation increases anthropomorphism
A summary of a taxonomy of anthropomorphic features in linguistic AI systems in the literature