Post on 11-Jan-2016
9/29/01 Human-Robot Interaction
Ecce Homo:
Why It’s Great to be Labeled a “Person”
Clifford NassStanford University
9/29/01 Human-Robot Interaction
Three Stories
• Robot best friend• Shylock
– Bleed– Tickle – Poison– Eyes, hands, organs, dimensions,
senses, affections, passions
• Dred Scott case
9/29/01 Human-Robot Interaction
Lessons to be Learned
• “Human” is nothing more than a label
• “Human”: not ontological• “Human”: psychological and social
9/29/01 Human-Robot Interaction
Does the Label Matter?
• Not philosophical question• Is experimental question
9/29/01 Human-Robot Interaction
Humor — Findings
– Funny people are:• Smarter• More likable• More friendly• Does not harm performance
– Even applies in task situations
9/29/01 Human-Robot Interaction
Humor — Variables
– Independent variables• Successful humor vs. no humor• HCI vs. CMC
– Dependent variables• Perception of interactant• Behavior• Smiling responses
9/29/01 Human-Robot Interaction
Likable
4
4.5
5
5.5
6
6.5
7
HCI CMC
HumorControl
9/29/01 Human-Robot Interaction
Cooperation
4
4.5
5
5.5
6
HCI CMC
HumorControl
9/29/01 Human-Robot Interaction
Actual Task Time
10
15
20
25
30
35
HCI CMC
HumorControl
9/29/01 Human-Robot Interaction
Joking
0
0.5
1
1.5
HCI CMC
HumorControl
9/29/01 Human-Robot Interaction
Smiling
0
0.5
1
1.5
2
2.5
HCI CMC
HumorControl
9/29/01 Human-Robot Interaction
Humor — Implications
• Profoundly human behaviors are acceptable on computers
• People will be verbally friendly to computers
• CMC is basically the same, but – More jokes– More smiles!
• Cognitive construction of person
9/29/01 Human-Robot Interaction
Adaptation: Findings
• Humans: Adaptation is universal• Computers: Adaptation is uncommon
– Hard problem– Focus on
• Technology• Precision of measurement
9/29/01 Human-Robot Interaction
Possible processes
• Social facilitation– Confident people do better when
observed– Non-confident people do worse
• Stress– Opposite of social facilitation
• Which explanation applies to:– human-computer interaction?– Human-human interaction?
9/29/01 Human-Robot Interaction
Method
• Computer-based GRE (same for everyone)
• IVs– Confidence in prior GRE performance
• High• Low
– Adaptation• No adaptation• Ostensibly adapts based on previous
questions
9/29/01 Human-Robot Interaction
Performance on the GRECMC
10.5
11
11.5
Confident Unconfident
FixedAdapting
9/29/01 Human-Robot Interaction
Performance on the GREHCI
8
10
12
Confident Unconfident
FixedAdapting
9/29/01 Human-Robot Interaction
Performance on the GRE
8
10
12
CMC HCI
Conf/ FixedConf/ AdaptingUnconf/ FixedUnconf/ Adapting
9/29/01 Human-Robot Interaction
Implications
• Human label influences response to adaptation– Social facilitation in CMC– Choking in HCI
• Rarity effects?
9/29/01 Human-Robot Interaction
Animated Representations
• Conformity pressure is powerful– Informative– Normative
• Appearance is powerful
9/29/01 Human-Robot Interaction
Method
• Choice-dilemma questionnaire• Representation
– Text box– Stick figure– Rich animated figure
• (Ostensible) Interactant– HCI– CMC
9/29/01 Human-Robot Interaction
Traditional CMC
9/29/01 Human-Robot Interaction
Animated “People”
9/29/01 Human-Robot Interaction
Intelligence of Agent
6
6.5
7
7.5
Trustworthy Competent
Stick Figure - CMC
Animated Figure - CMC
Stick Figure - HCI
Animated Figure - HCI
9/29/01 Human-Robot Interaction
Public Conformity
3.5
4.5
5.5
CMC HCI
One Actor - Public
Four Actors - Public
One Actor - Private
Four Actors - Private
9/29/01 Human-Robot Interaction
Implications
• Representation matters in HCI• Representation matters in CMC!
– Was that the mental construction?• TTS evidence
– Can’t draw general conclusions about faces, characters, etc.
• CMC exerts great social influence
9/29/01 Human-Robot Interaction
Visible Agents
• What happens when agents are indistinguishable from humans?– Distinction disappears or– Label makes a difference
• Expectancy Theory: Everything should match its label
9/29/01 Human-Robot Interaction
Labeling: Findings
• Expectation of consistency with label– Better than expected: Great!– Worse than expected: Awful!
9/29/01 Human-Robot Interaction
Demo
• Human Slide51.human.movor
• Synthetic Slide51.synthetic.mov
9/29/01 Human-Robot Interaction
Context
• Auction Site• Human-looking or agent-looking• “Human” or “Agent”
9/29/01 Human-Robot Interaction
Bidding Behavior
50
60
70
Human Agent
HumanAppearanceAgentAppearance
9/29/01 Human-Robot Interaction
Implications
• Labels matter• Appearance matters• Humans are held to higher
standard
9/29/01 Human-Robot Interaction
Summary – Part I
• Turning Turing Test on its head– “Human” is not ontological– Not “Is it human” but “if it’s human, now
what?”
• HCI does NOT equal social interaction• But what about “Media Equation”?
– Cognitive construction may be at the critical determinant
9/29/01 Human-Robot Interaction
Final Question
• Will robots ever be labeled “human”?