interactions agents and social creativity,
TRANSCRIPT
![Page 1: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/1.jpg)
Social Creativity,Agents andInteractions
Simo Linkola(simo.linkola at helsinki.fi)
8.4.19
![Page 2: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/2.jpg)
Example: Robotic InstallationVideo 1 (3:32)
Video 2 (4:18) (no speech)
![Page 3: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/3.jpg)
Example: Robotic InstallationVideo 1 (3:32)
Video 2 (4:18) (no speech)
Which perspectives are present?
![Page 4: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/4.jpg)
Outline1. Sociality in Creative Systems2. Multi-agent Systems in CC3. Modeling the Environment4. Systems View of Creativity5. Interaction and Agent Roles6. Examples of Creative Social Agents
![Page 5: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/5.jpg)
Sociality in Creative Systems
![Page 6: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/6.jpg)
Why (Computational) Social Creativity?
● “Creativity does not happen in a vacuum”● Interaction is essential for human creativity
○ Learning domain knowledge and skills○ Cultural influence, etc.
● Interaction allows the system to potentially adjust its creative process during its lifetime○ Transformational creativity
![Page 7: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/7.jpg)
Sociality in Creative Systems
A single agent
Thousandsof agents
![Page 8: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/8.jpg)
Sociality in Creative Systems
A single agent
Thousandsof agents
● Elaborate design and interaction
● Real creative responsibilities
● Focus on a single agent behavior (or particular algorithms)
![Page 9: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/9.jpg)
Sociality in Creative Systems
A single agent
Thousandsof agents
● Elaborate design and interaction
● Real creative responsibilities
● Focus on a single agent behavior (or particular algorithms)
● Simple design and interaction
● Generation routines, minimal creativity
● Focus on emergent behavior
![Page 10: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/10.jpg)
Sociality in Creative Systems
A single agent
Thousandsof agents
Computational Social Creativity
● Computational social creativity:○ Research interests mainly in super-individual properties (emergence)○ What are the individual properties and interaction types that cause certain phenomena?
(proof-of-concepts)● In this lecture we discuss both:
○ Computational social creativity○ CC with a twist of (multiple) interacting agents
![Page 11: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/11.jpg)
Sociality in Creative Systems
A single agent
Thousandsof agents
Computational Social Creativity
Multi-agent Systems Agent-based Models
![Page 12: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/12.jpg)
Computational Social Creativity (CSC)1. Models should demonstrate a mechanism
a. Qualitative analysis
2. Models should be simple and reproducible3. Models should preferably generate new hypotheses
Saunders and Bown - Computational Social Creativity (2015)
![Page 13: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/13.jpg)
![Page 14: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/14.jpg)
Multi-agent Systems in CC
![Page 15: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/15.jpg)
Multi-agent Systems● Agents are autonomous and have local views
○ Agents take some individual responsibilities in the system○ Agents can only “see” part of the system at any given time
● No central controlling agent● There is always an environment, however, its role might be negligible● Interaction between agents (either direct or indirect) is a central
concept● Two ends of a spectrum:
○ A lot of simple agents, focus on emergent behavior○ A small set of intelligent agents, focus on fulfilling specific tasks
![Page 16: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/16.jpg)
Characteristics of MAS in CC● Mostly software agents
○ Some work on robotics
● Stationary societies● Agents create artifacts in some domain(s)● Agents have a way to evaluate artifacts● Interaction is mostly direct
○ Agents exchange messages directly with each other○ Most typical message is artifact exchange: other agent evaluates its peer’s artifact
![Page 17: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/17.jpg)
Nwana’s Software Agent TopologyCOOPERATE
LEARNAUTONOMOUS
SMART AGENTS
COLLABORATIVE LEARNING AGENTS
COLLABORATIVE AGENTS
INTERFACE AGENTS
Nwana - Software Agents: An Overview (1996)
![Page 18: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/18.jpg)
Stationary Agent Societies● Not really an open environment:
○ The designer chooses agents, their communication models, etc.
● Agents are (somewhat) homogenous● Agents do not leave or enter after initialization● Agents are generally assumed to be “well behaving”● Agents may have connections to other agents at the initialization time
○ Even knowledge of all other agents
● In many cases, societies are run in iterative simulations○ Useful for studying specific interests by simplifying the model
![Page 19: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/19.jpg)
Artifacts● Artifacts can be thought as solutions to a (search) problem (CSF)● Examples:
○ A piece of text generated with a Markov chain○ An image created with genetic programming○ A mathematical theory created via concept blending○ Non-photorealistic rendering of an image
![Page 20: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/20.jpg)
Evaluation● Agents have some means to evaluate the artifacts they (and others)
create● E.g. Extract features from an artifact (and combine them)● In ideal case, evaluation takes all the following into account:
○ Is an artifact novel?○ Is an artifact of high value?○ Is an artifact surprising?○ Intention of the artifact?
● The situation and agent’s traits affect evaluation○ E.g. agent’s memory, aesthetic preferences
● Designing good evaluation methods is difficult!
![Page 21: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/21.jpg)
Evaluation: Examples
![Page 22: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/22.jpg)
Goal of the Agents● Cooperative
○ All agents have the same goal, or○ Agents benefit from fulfilling other agents’ goals○ Agents can also be called altruistic
● Competitive○ All agents have their own goal they try to satisfy○ Fulfilling a goal may render some other goals impossible○ Agents can also be called selfish
● Mixture○ Both cooperative and competitive goals
● Intrinsic Motivation○ See e.g. Schmidhuber. 2010. Formal Theory of Creativity, Fun and Intrinsic Motivation
![Page 23: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/23.jpg)
Modeling the Environment
![Page 24: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/24.jpg)
Modeling the Environment● Environment: agents and/or other (static) context● Why model the environment? Suggestions?
![Page 25: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/25.jpg)
Modeling the Environment● Environment: agents and/or other (static) context● Why model the environment?
○ A way for an agent to reason about its choices○ Otherwise interaction happens by chance/randomly○ Without modeling, the agent cannot adapt to different situations
![Page 26: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/26.jpg)
Modeling the Environment● Environment: agents and/or other (static) context● Why model the environment?
○ A way for an agent to reason about its choices○ Otherwise interaction happens by chance/randomly○ Without modeling, the agent cannot adapt to different situations
● Modeling the environment allows the agent to gain interaction awareness○ Reflection and control of the interactions
![Page 27: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/27.jpg)
Modeling the Environment● Most common approach is to use reinforcement learning● Tries to maximize expected utility (reward) in the long run● Selects “optimal” action in each situation (state)● E.g. Q-learning
![Page 28: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/28.jpg)
Modeling the Environment
SELECT ACTION WITH THE HIGHEST EXPECTED UTILITY
Plan next action
![Page 29: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/29.jpg)
Modeling the Environment
Execute action
EXECUTE THE PLANNED ACTION
Plan next action
![Page 30: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/30.jpg)
Modeling the Environment
Plan next action
Execute action
Perceive environment
OBSERVE UTILITY OR VALUEOF THE ACTION
![Page 31: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/31.jpg)
Modeling the Environment
Update model
UPDATE MODEL BASED ON THE OBSERVED UTILITY
Plan next action
Execute action
Perceive environment
![Page 32: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/32.jpg)
Modeling the Environment
Update model
START FROM THE BEGINNING
Plan next action
Execute action
Perceive environment
![Page 33: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/33.jpg)
The Systems View of CreativityAlso known as Domain-Individual-Field Interaction (DIFI model)
![Page 34: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/34.jpg)
The Systems View of Creativity
SOCIETY
CULTURE
PERSONAL BACKGROUND
FIELD INDIVIDUAL
DOMAINSelects novelty
Transmits information
Produces novelty
Stimulates novelty
Csikszentmihalyi. 1988. Society, Culture and Person: The Systems View of Creativity
![Page 35: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/35.jpg)
The Systems View of Creativity
SOCIETY
CULTURE
PERSONAL BACKGROUND
FIELD INDIVIDUAL
DOMAINSelects novelty
Transmits information
Produces novelty
Stimulates novelty
Repository of cultural knowledge/information
![Page 36: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/36.jpg)
The Systems View of Creativity
SOCIETY
CULTURE
PERSONAL BACKGROUND
FIELD INDIVIDUAL
DOMAINSelects novelty
Transmits information
Produces novelty
Stimulates novelty
Repository of cultural knowledge/information
Brings about transformations of to the domain knowledge
![Page 37: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/37.jpg)
The Systems View of Creativity
SOCIETY
CULTURE
PERSONAL BACKGROUND
FIELD INDIVIDUAL
DOMAINSelects novelty
Transmits information
Produces novelty
Stimulates novelty
Repository of cultural knowledge/information
Brings about transformations of to the domain knowledge
Selects the knowledge that is worth preserving
![Page 38: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/38.jpg)
The Systems View of Creativity
SOCIETY
CULTURE
PERSONAL BACKGROUND
FIELD INDIVIDUAL
DOMAINSelects novelty
Transmits information
Produces novelty
Stimulates novelty
Relation to Boden’s personal creativity and historical creativity?
![Page 39: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/39.jpg)
The Systems View of Creativity
SOCIETY
CULTURE
PERSONAL BACKGROUND
FIELD INDIVIDUAL
DOMAINSelects novelty
Transmits information
Produces novelty
Stimulates novelty
Produces (personal) creativity
Selects historical creativity
Relation to Boden’s personal creativity and historical creativity?
![Page 40: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/40.jpg)
Interaction and Agent Roles
![Page 41: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/41.jpg)
Interaction● Agents interact with each other one way or another● We are mostly interested in direct interaction● Some specific interaction types:
○ Negotiation○ Voting○ Collaboration
● Emerging social phenomena:○ “Best friends”○ Cliques○ Gatekeepers○ Other social structures
![Page 42: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/42.jpg)
Interaction● Communication model can be nearly anything:
○ Tell a friend○ Report to a supervisor○ Broadcast to peers○ Write on a blackboard
● Interest in models which restrict communication○ Broadcasting everything causes bandwidth problems○ Agents are modeled to have limited resources○ Interaction awareness one of the goals
■ Model the interaction and choose how to interact
![Page 43: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/43.jpg)
Negotiation● Agents negotiate over:
○ a contract○ the terms of transactions○ etc.
● Typical in Consumer - Producer models, e.g. e-commerce● May happen between only two agents or a set of agents● Typically iterative
For more information see, e.g. Lopes, Woolridge and Novais. Negotiation among autonomous computational agents: principles, analysis and challenges. 2008. Artificial Intelligence Review.
![Page 44: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/44.jpg)
Negotiation: ExampleAgent A: Give me service X.
Agent B: I’ll give you service X if you give me service Y.
Agent A: Give me service X and I’ll give you service Z.
… etc.
![Page 45: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/45.jpg)
Voting● A set of agents A● A set of candidates C● Each agent in A votes by returning C sorted by preference● Selected voting rule defines which candidate(s) get selected● Voting rules:
○ Majority, Condorcet○ Dictatorship○ Run-off rules○ Etc.
● No silver bullet, have to pick the “least evil” option (voting rule)● Example: Agents vote for an artifact from a set of artifacts
![Page 46: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/46.jpg)
Collaboration● Special interest in creativity: brainstorming, game design, etc.● Collaborating agents:
○ May produce more value and/or novelty○ May be able to produce artifacts they could not alone
● Benefits from agents with reasonably different properties● Agents need to be able to agree on critical decisions● Often domain and generator dependent
○ Agents should share artifact representations or be able to translate them○ What is communicated during the collaboration?
![Page 47: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/47.jpg)
Emerging Social Phenomena● Interest in computational social creativity● Cliques / coalition formation
○ Agents form tight social circles where they like artifacts produced by the agents in their circle more than agents outside the circle.
● Gatekeepers○ Some agents form a role of a “gatekeeper”, where they have substantially more control
on the artifacts that are accepted to the domain than other agents.
● Etc.
![Page 48: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/48.jpg)
Agent Roles in Computational Creativity● Designers-Clients (Zhang and Saunders)
○ Designers are producers which create new designs (artifacts)○ Clients are consumers which accept designer’s design
■ Clients can also give requirements for the designers
● Critics○ Only evaluate artifacts, can be seen also as consumers○ Should have sophisticated evaluation standard
● Gatekeepers (e.g. Sosa and Gero)○ Individuals with high influence on the artifacts that are accepted to the domain○ Emergent role forming as an aggregate result of social interactions
● The roles depend on what is modelled on each occasion
Zhang and Saunders. Exploring Conceptual Space in Language Games Using Hedonic Functions (2014)Sosa and Gero. Social Models of Creativity (2005)
![Page 49: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/49.jpg)
Examples of Creative Social Agents
![Page 50: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/50.jpg)
Evolution of Culture (EVOC) by Liane Gabora
● Society of neural network based agents in a 2D grid-world:○ Neural network encodes ideas for actions and detects trends of fit actions○ Body implements the actions○ Agent’s goal is to attract a mate and make tools (two fitness functions)
● Cultural phenomena:○ Imitation: agents copy neighbours actions, allows sharing of effective actions○ Invention: agents can modify their previous actions○ Knowledge-based operations: new ideas are generated strategically (i.e. not random)○ Mental stimulation: before implementing an idea, the agent simulates its usage
● Interest in diversity of actions in the population as a function of time
Gabora has published many articles around the same concept. See, e.g. Gabora, L. (2008). EVOC: A computer model of the evolution of culture. In V. Sloutsky, B. Love & K. McRae (Eds.), 30th Annual Meeting of the Cognitive Science Society. Washington DC, July 23-26, North Salt Lake, UT: Sheridan Publishing.
![Page 51: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/51.jpg)
Curious Design Agents by Rob Saunders and John Gero
● Agents aim to produce novel artifacts (spirographs)● Agents have a model (SOM) of previously seen artifacts● Agents evaluate the novelty of an artifact w.r.t the model● Agents have different levels of preferred novelty● Hedonic function follows the Wundt Curve● Agents may give positive feedback of artifacts● Emergence of cliques is observed
○ Agents that give positive feedback to each other○ Agents in a clique have the same level of preferred novelty
See, e.g. Rob Saunders and John S. Gero. 2002. How to study artificial creativity. In Proceedings of the 4th conference on Creativity & cognition (C&C '02). ACM, New York, NY, USA, 80-87.
![Page 52: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/52.jpg)
Novelty-seeking Agents by Linkola et al.
● Generated also spirographs as Saunders and Gero● Agent memory was a list of previously seen artifacts● All agents aimed to maximize novelty● Studied how (social) decision-making affects the novelty of the produced
artifacts○ Voting and self-criticism & veto capabilities
● Voting is effective in selecting the most novel artifacts● Veto and self-criticism effectively constrain the number of artifacts subject
to voting● An agent society with locally informed agents can outperform fully
informed monolithic systemLinkola et al. 2016. Novelty-seeking Multi-agent Systems. In Proceedings of the 7th International Conference on Computational Creativity. Paris, France.
![Page 53: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/53.jpg)
Reflexive Looper by Sony CSL
● Part of Flow Machines project lead by François Pachet● Musical companion which adapts to you● Plays chords, drums, bass, etc.● Video (4:09)
![Page 54: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/54.jpg)
Social Models of Creativity by Ricardo Sosa and John Gero
● Study creativity in the DIFI-setting (Domain-Individual-Field-Interaction)● Agents make designs in simple two-dimensional shapes● Designers and adopters● Findings about many individual and social aspects of creativity, e.g.
○ Change agents: trigger change cycles in a bottom-up direction○ Gatekeeping: some adopters become gatekeepers by an aggregate result of social
interaction
![Page 55: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/55.jpg)
Collaborator Selection for Creative Agents by Hantula & Linkola
● Study how agents may model their peers and use the learned peer models to select favourable collaboration partners
● Agents create images using GP○ Agents can collaborate and create image as a pair
● Agents have different aesthetic measurements and they can change their evaluation during the simulation
○ Peer model needs to be dynamic
● Peer models are beneficial in choosing the collaboration partners○ may aid selecting even agents with different aesthetic measurements
Linkola et al. 2018. Towards Goal-aware Collaboration in Artistic Agent Societies. In Proceedings of the 9th International Conference on Computational Creativity. Salamanca, Spain.
![Page 56: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/56.jpg)
![Page 57: Interactions Agents and Social Creativity,](https://reader030.vdocuments.us/reader030/viewer/2022012712/61ab4713bf428819ed3d3166/html5/thumbnails/57.jpg)
Summary● Creativity does not happen in a vacuum
○ System’s view of creativity as a conceptual framework
● Interaction is an important aspect in many creative systems○ Direct and indirect interaction are both valid, focus on direct
● Cooperation and collaboration○ Cooperation may produce results which agents are not able to produce alone
● A truly autonomous agent needs to have (learn) a model of its environment to be able to act intentionally
○ Interaction awareness!