week 5 - lecture 11 Flashcards
characteristics of social artificial agents
- social presence (interactive potential, language and conversation, social affordances, identity)
- sentience (motivation, empathy, cognition, subjectivity, drive to bond)
- agency (ability to sense the environment and act upon it, self control, decision making, autonomy)
embodiment (material presence, physical affordances, appearance, likeness) - purpose & functions (assistve, educational, social companionship, hospitality, service & entertainment)
- embodiement
how to build social artificial agents
- user’s perspective
- perceived sentience, agency, social presence
reaction (cognitive-effective)
- engagement & acceptance
- perceived realism
- subjective relevance - design features
- embodiment
- features
- signaled sentience, agency, social presence - context
- audience demographics
- interaction goal
- environment
- ethics & legality
4 types of involvements of robots
- factory & industrial robots
- task-driven control & supervision
- assembly robots
- mining robots - cobots (collaborative robots)
- task-driven interaction
- surgical robots
- engineering robots - social robots
- relationship-driven interaction
- robot pets - social assistive robots
- task & relationship-driven interaction
- robot tutors
- robot caregivers
3 different types of application fields
- healthcare
- education
- hopsitality, service and entertainment
roles, frequent target users, value & benefits and concerns of healthcare robots
roles: assistive, monitoring, companionship and well-being and health self-management
frequent target users: children, elderly and clinical populations
value & benefits
- mitigate personnel & resources shortage
- increase level of attention
- promote treatment adherence
- associated with improvements in health and well being
concerns
- fear of loss of human contact
- infantilization
- deception & privacy
roles, frequent target users, value & benefits and concerns of educational robots
roles: tutors & co-teachers, peers and learners
frequent target users: children, adults
values and benefits
- mitigate personnel & resources shortage
- facilitate classroom management
- provide alternative learning experiences
- promote engagement & enjoyment
- improve learning outcomes
concerns
- fear of loss of human contact
- artificial agents handle unpredictability and chaos poorly
- physical hazards (malfunction or abuse)
- enabling undesirable behavior (bullying)
- agent inflexibility not suitable for education
- toy or educational device?
roles, frequent target users, value & benefits and concerns of hospitality, service and entertainment robots
roles: host & guidance, customer service, waiters
target users: anyone
value and benefits:
- in-between machine
- novelty & user engagement
- can promote self-disclosure, increase sense of intimacy and connectedness
- can get away with violations to social rules
concerns
- fear of loss of human contact
- artificial agents handle unpredictability and chaos poorly
- physical hards (malfunction or abuse)
- can get away with violations to social rules
bias in decision making systems
biased, unfair predictive data analytics in business, healthcare can erode public liberties and rights
decision making process opaque (niet duidelijk), no accountability
bias can be learned from programmer
the 5 practical concerns artificial agents
- privacy and surveillance
- AI facilitates collection of analysis of personal, sensitive data
- social interaction with artificial agents as additional data source: personal, sensitive facts can be released to social agents - lack of transparency
- machine learning and algorithmic principles opaque to designers, programmers and users - automation & employment
- productivity gains → fewer humans are required for the same output - legality
social and civil rights for artificial agents?
laws and regulations for human artificial intelligence - physical hazards
psychosocial risk factors
agent-caused injuries and damages
the problem of honesty vs realism of social artificial agents
- humans attribute mental properties to objects, and empathize with them, especially when the outer appearance looks like real humans
- this can be used to deceive humans intro attributing more intellectual or even emotional significance to robots or AI systems than they deserve
- few constraints of ethics and law apply to robot
3 field were you could discuss if robots are better than humans
- military and kill-bots
- relationship and sex-bots
- human agents LTRs
pros, cons and main concerns of military and kill-bots
pros
- people don’t need to fight anymore
- military robots might be more accurate, causing less c- ollateral damage
increased defensive capabilities without compromising population
- reduces costs and effort
cons
- robots not constrained by compassion or empathy
- trivializes killing and war
- robots deal poorly with chaos and ambiguity
main concerns
- a steady increase in the intergotian of AI in military systems is likely
- arms-race: significant international competition in military AI. Those risks are:
- ethical: humanitarion implications
- strategic: increase war likelihood, conflict escalation and malicious actor proliferation
operational: reliability, fragility and security of AI systems
pros, cons and main concerns of relationship and sex-bots
pros
- can satisfy social, affective and physical needs not otherwise fulfilled
- provide an outlet for socially unacceptable or illegal tendencies
- could provide a safe therapeutic environment
- entertainment and novulty
cons
- might encourage acting out unacceptable, illegal tendencies
- trivialized intimacy and human connection
- might provide a ‘sandbox’ environment to train unacceptable, illegal practices
- may lead to physchological damage
main concerns
- a steady increase in the development, sophistication is ongoing
- market-race: high consumer deman, and it’s increasing
- potential to promote and normalize psychological and behavioral deviations
- potential to discourage/replace human-human interaction
compromise notions of consent, aesthetics and social exchange
pros, cons and main concerns of human-agents LTRs
pros
- can satisfy needs not otherwise fulfilled
- support non-traditional relationship preferences
- alternative for those not satisfied with human-human relationships
cons
- not a real relationship (unbalanced)
- trivializes relationships
- loss of humanity and human connection
- deterioration of social and interpersonal skills
main concerns
- amplify and normalize anti-social tendencies
- market-race: high consumer demand, and it’s increasing
- potential to promote and normalize psychological and behavioral deviations
- potential to discourage/ replace human-human interaction
- social impact: reduced birth-rates, socialization and community building
2 ways that create toxic relationships
- abuse & exploitation
- deception and manipulation
