Task 2 Flashcards
What is creativity, and how does it differ across domains?
Creativity varies across domains (e.g., sports, arts, science), but all forms share common aspects such as novelty, innovation, and problem-solving.
What is Computational Creativity (CC)?
CC is an AI field that studies how computers can generate creative outputs by simulating human-like creative processes through algorithms and data structures.
What are some key questions that Computational Creativity aims to answer?
Where does creativity reside (process, product, or creator)?
How does creativity relate to expertise and domain knowledge?
How is creativity judged and measured?
How do collective behaviors contribute to creativity?
What are the four key criteria that define creative solutions?
Novelty & Usefulness – The solution must be new and beneficial.
Rejection of Previous Ideas – It must break conventional thought patterns.
Motivation & Persistence – It requires intense effort and perseverance.
Problem Clarification – It often transforms a vague problem into a clear one.
What is the Investment Theory of Creativity?
This theory suggests that a creative agent (human or machine) must recognize and justify the unexpected value of its outputs, rather than just generating content randomly.
Why is intentionality important in creativity?
Mere generation of outputs (like AI producing random images) is not enough; creativity requires awareness of quality, novelty, and impact.
What is pastiche in computational creativity?
Pastiche is the imitation of artistic styles without true originality. AI-generated art often falls into this category, as it replicates patterns without innovating.
What is the goal of CC beyond pastiche?
To develop AI that transcends imitation and demonstrates human-like creativity by taking risks, self-critiquing, and learning from failures.
What is exploratory creativity?
It involves searching within an existing conceptual space, finding new pathways or unconventional connections (e.g., a chess player inventing a new opening strategy).
What is transformational creativity?
It alters the conceptual space itself, leading to paradigm shifts (e.g., Newtonian physics to Einstein’s relativity).
What components make up a neural network in AI?
Processing units (nodes) – Simulate neurons.
Activation states – Nodes activate when they cross a threshold.
Connections between nodes – Can be excitatory or inhibitory.
Input/Output rules – Define how nodes process information.
Learning rules – Like Hebbian learning, where co-activated nodes strengthen their connections.
How do neural networks store and retrieve information?
They act as content-addressable memories, retrieving patterns based on partial inputs (e.g., seeing part of a cat’s face activates the “cat” concept).
What are the four stages of creativity?
Preparation – Gathering knowledge & relevant information.
Incubation – The problem is set aside, allowing subconscious processing.
Illumination – The “aha” moment when a solution appears.
Verification – Refining and testing the idea for feasibility.
How does attention affect creativity?
Focused Attention – Helps analyze details but limits new ideas.
Defocused Attention – Allows more remote associations, increasing creativity.
What is the relationship between intelligence and creativity?
Creativity correlates with intelligence only up to an IQ of 120, beyond which the correlation disappears.
What brain networks support creative thought?
Default Mode Network (DMN) – Generates spontaneous, self-generated thoughts (e.g., imagination, mind-wandering).
Control Network (CN) – Evaluates and refines creative ideas through cognitive control.
Salience Network (SN) – Helps switch between idea generation (DMN) and idea evaluation (CN).
How do these networks interact during creativity?
During idea generation, DMN is more active.
During evaluation & refinement, CN becomes more involved.
Highly creative individuals show greater connectivity between DMN and CN.
Why does AI struggle with true creativity?
AI lacks:
Intentionality (it does not “care” about its outputs).
Emotional reasoning (it cannot weigh moral, aesthetic, or emotional value).
Independent thought (it depends on human-defined inputs).
What is Creativity 4.0?
A new model where AI, human programmers, expert evaluators, and domain knowledge interact to drive AI-assisted creativity.
How does evolutionary computation contribute to AI creativity?
It simulates natural selection using:
Random variations (mutation) – Generates new possibilities.
Selective retention – Keeps the most promising solutions.
Crossover (recombination) – Mixes elements from different solutions.
Unlike traditional AI that hill-climbs toward a known solution, evolutionary computation allows:
Exploration of a vast solution space (parallel searches).
Discovery of unexpected solutions (avoids local optima).
Adaptability to complex creative tasks (mimicking human insight).
How can AI creativity evolve in the future?
AI must:
Move beyond pastiche and learn to assess value.
Integrate more human-like cognitive flexibility (e.g., balancing divergent and convergent thinking).
Improve self-awareness & evaluation mechanisms to filter outputs meaningfully.
What are the main perspectives on where creativity resides?
Creativity can be found in:
The creator (personality, intelligence, motivation).
The process (mental steps, cognitive mechanisms).
The product (originality, usefulness, impact).
A combination of all three.
How does creativity relate to expertise?
Expertise provides foundational knowledge, but excessive expertise can sometimes limit creativity by reinforcing conventional thinking.
What is the goal of Computational Creativity (CC)?
To create AI systems that:
Generate novel and useful outputs
Assess the value of their creations
Improve their creativity through feedback and learning
How does CC differ from traditional AI?
Traditional AI focuses on problem-solving using structured rules, while CC seeks to produce original and unexpected ideas.
How do exploratory and transformational creativity differ?
Exploratory Creativity – Searches within an existing space (e.g., remixing musical elements).
Transformational Creativity – Changes the space itself, redefining what is possible (e.g., Einstein’s shift from Newtonian to relativistic physics).
What are the different levels of creativity?
Little-C Creativity – Everyday creativity (e.g., trying a new recipe).
Pro-C Creativity – Professional creativity (e.g., an engineer solving a novel design problem).
Big-C Creativity – Groundbreaking contributions that redefine a field (e.g., Picasso, Einstein).
What is Campbell’s BVSR theory of creativity?
Creativity results from random idea generation (blind variation) followed by selecting and keeping useful ideas (selective retention).
What are primary and secondary process thinking in creativity?
Primary Process Thinking – Associative, imaginative, non-logical (useful for idea generation).
Secondary Process Thinking – Logical, structured, reality-based (useful for refining and implementing ideas)
How does evolutionary computation enable AI creativity?
It mimics natural selection:
Mutation – AI randomly alters solutions.
Crossover – AI combines elements of past solutions.
Selection – AI keeps the most successful results.
What is simulated annealing, and how does it relate to creativity?
Simulated annealing is an AI technique that prevents getting stuck in local optima by occasionally exploring more random, less obvious solutions—similar to how humans have creative breakthroughs after periods of mental wandering.
Why are jokes considered a form of creative cognition?
Jokes subvert expectations, using unexpected shifts in logic to create humor—similar to how creativity involves making unconventional connections.
How do contradictions contribute to creativity?
While AI sees contradictions as dead ends, creativity treats them as opportunities for deeper exploration (e.g., paradoxes in scientific discoveries).
What are some ethical concerns about AI creativity?
Bias in AI-generated content – AI can reflect human prejudices.
Authorship issues – Who owns AI-generated art?
Moral decision-making – AI lacks ethical reasoning beyond simple heuristics.
What is a Hopfield Network?
A type of recurrent neural network where every node is connected to every other node, and it evolves to minimize its energy function.
What are the key properties of a Hopfield Network?
Fully connected – Each node is linked to every other node.
Symmetric weights – The connection strength from node A to B is the same as from B to A.
Binary activation – Nodes take values of +1 or -1.
Energy minimization – The network evolves towards stable low-energy states (fixed-point attractors).
What happens when a Hopfield Network runs?
The network updates one node at a time, adjusting its state to minimize the overall energy of the system
What are attractors in a Hopfield Network?
Stable energy-minimized states where the system settles, functioning as memory storage.
What problem can occur in energy minimization?
The network can get stuck in local minima, where the energy is reduced but not at its lowest possible state (global minimum)
What is simulated annealing in Hopfield Networks?
A method to escape local minima by introducing randomness (high “temperature”) and gradually lowering it to refine solutions.
How does temperature affect activation in a Hopfield Network?
Low temperature → Deterministic behavior, nodes follow strict activation rules.
High temperature → Nodes behave randomly, helping escape local minima.
How is simulated annealing linked to creativity?
High temperature (~low cortical arousal) allows free associations (like brainstorming).
Low temperature (~focused attention) enables refinement and evaluation of ideas.
The brain may use similar processes in problem-solving and creativity.
How can Hopfield Networks be used to model creative thinking?
They simulate the alternation between idea generation (randomness) and refinement (structured thinking), similar to human creativity.
How do Hopfield Networks relate to cortical arousal?
High cortical arousal corresponds to focused, structured thinking, while low cortical arousal corresponds to free-associative, creative thinking.
What is the cortical scanning hypothesis?
The brain’s arousal system fluctuates in cycles, possibly simulating periodic annealing to help problem-solving and creativity.
How might the brain avoid getting “stuck” in a local minimum?
By shifting between different levels of arousal and focus, similar to how simulated annealing introduces randomness before refining solutions.
