5. PROBLEM:SIMULATINGROUTINES
Humans make habitual decisions, interconnect these habits throughout the day and
use these interconnected habits as a blueprint for social interaction.
you intentionally go to the
office to trigger the routine
of working
You colleague waits till the
coffee break to discuss current
matters
I’ll read a paper now (which
connects to working) and drink
coffee later (which connects to
taking a break
Activate
work habit!
Connect
activities! Expect similarity!
9. TheMainDifficulty
The difficulty in modelling a domain-independent framework is to
balance reuseability with grounding in empirical work.
Reuseable Grounded
in
Emperical
Work
Generality
Specifity
Vague literature describing
social phenomena
Concrete requirements
Verifying whether these
requirements hold in the model
Concepts that
generalize over domains
u
Modular
Compact
10. Overview
The dissertation describes (I) models for agents with values and norms
(II) the SoPrA agent framework (III) applications of the framework
I.Agents with
Values and
Norms
II. Agent
Framework
forHuman
Routines
III.
Applications
12. CreatingSoPRA
Social practices – human routines – integrate three aspects of human
behaviour: habituality, sociality and interconnectedness.
II.Agent
Framework
for Human
Routines
13. CreatingSoPRA
Social practices – human routines – integrate three aspects of human
behaviour: habituality, sociality and interconnectedness.
II.Agent
Framework
for Human
Routines
14. CreatingSoPRA
Social practices – human routines – integrate three aspects of human
behaviour: habituality, sociality and interconnectedness.
II.Agent
Framework
for Human
Routines
15. CreatingSoPRA
Social practices – human routines – integrate three aspects of human
behaviour: habituality, sociality and interconnectedness.
II.Agent
Framework
for Human
Routines
16. CreatingSoPRA
Social practices – human routines – integrate three aspects of human
behaviour: habituality, sociality and interconnectedness.
Activate
work habit!
Connect
activities! Expect similarity!
II.Agent
Framework
for Human
Routines
17. Applications
SoPrA is used to support simulations on meat-eating, AI, hospitals, and
habitual theories.
Simulating Meat
Eating
Aligning Values in AI Identifying Social
Bottlenecks in
Hospitals
Discerning Two
Theories on Habits
III. Applica-
tions
22. IdentifyingSocialBottlenecksinHospitals
SoPrA helps in identifying wether the staff can help each other out in a
given (reproduced) emergency room.
2. A head nurse can
cover some of the
necessary tasks of the
secretary.
3. The staff can help each
other out, because they
know the equipment the
others need.
III. Applica-
tions
23. Overview
This results in an evidence-driven framework useable for a wide range of
simulation studies that involve human routines.
I.Agents with
Values and
Norms
II. Agent
Framework
forHuman
Routines
III.
Applications
24. DiscerningTwo TheoriesonHabits
Theory A states that non-activitated habits weaken while theory B states
that non-activated habits stay the same.
III. Applica-
tions
25. DiscerningTwo TheoriesonHabits
In a certain scenario we found that the two theories should lead to
different observations: this helps to set-up an empirical experiment.
III. Applica-
tions
Reuseability
Not using the same concepts twice in some different form
Making sure there is no overlap
That you split up the model nicely
This results in an agent framework with a clear relation to current evidence and, due to its modularity and focus on domain-independence, is usable for a wide range of ABS studies that involve human routines. As such, SoPrA is relevant for scientific work in (1) ABS by enabling a new way to know, explore and improve the world, grounded in evidence on human routines; (2) in multi-agent systems by enabling agents that understand and interact with human routines; and (3) in the social sciences by crystallizing theories on human routines and enabling exploration of these theories via simulation. Furthermore, this thesis shows the societal relevance of SoPrA for understanding and improving the role of routines in AI safety, emergency rooms, commuting behaviour and consumption behaviour.
Values are what we find important in life
Norms are standards society
They are both very interesting and evidence-driven but do not suffice
A main insight was that we could model these three similar dimensions
Interconnectedness: your activity to go to this ceremony is similar to your activity of going back
An more combinations of these
A main insight was that we could model these three similar dimensions
An more combinations of these
Main difficulty: translating theory to concreate computational models
Main difficulty: translating theory to concreate computational models
Hoe mijn raamwerk gedrag modelleert: hoe eetgewoontes veranderen over de tijd; je ziet de kleurtjes veranderen
Elk puntje is een person
Elk blokje is een locatie
Main difficulty: translating theory to concreate computational models
Main difficulty: translating theory to concreate computational models
Main difficulty: translating theory to concreate computational models
Main difficulty: translating theory to concreate computational models
This results in an agent framework with a clear relation to current evidence and, due to its modularity and focus on domain-independence, is usable for a wide range of ABS studies that involve human routines. As such, SoPrA is relevant for scientific work in (1) ABS by enabling a new way to know, explore and improve the world, grounded in evidence on human routines; (2) in multi-agent systems by enabling agents that understand and interact with human routines; and (3) in the social sciences by crystallizing theories on human routines and enabling exploration of these theories via simulation. Furthermore, this thesis shows the societal relevance of SoPrA for understanding and improving the role of routines in AI safety, emergency rooms, commuting behaviour and consumption behaviour.
Main difficulty: translating theory to concreate computational models
Main difficulty: translating theory to concreate computational models
When compared to a ‘R’ a lot of detail is represented based on the agent itself and its behaviour
Heterogeneity: agents differ in if they are sick or not but also in age, and they are individually represented
Agents interact based on proximity here
Emergence we see some organization of how the virus spreads when we zoom out that is difficult to predict by only looking at one individual