Overview of What Will A.I. Do to (and for) Human Creativity?
This first part of a two-episode Freakonomics Radio series asks whether AI will diminish, replace, or amplify human creativity. Stephen Dubner explores the idea that AI may become an intimate “exo-self” — an always-on creative partner that knows you well enough to advise, guide, and even collaborate with you — while also highlighting the real risks to creative jobs, originality, and ownership. The episode balances optimism about AI as a creative tool with skepticism about its cultural and economic consequences.
The Big Idea: AI as an “Exo-Self”
Kevin Kelly, co-founder of Wired, introduces the concept of an exo-self: an AI agent embedded in your phone or smart glasses that is always on, always observing, and increasingly attuned to your preferences and habits.
Kelly’s four possible relationships with AI
He imagines AI acting as:
- Advisor — a personalized source of guidance
- Guardian/parent — protective, wise, and risk-aware
- Friend/hero — supportive, encouraging, even role-model-like
- Twin/clone — a near-identical collaborator that thinks and reacts like you
Kelly sees the twin/cloned version as especially powerful, because it could give people speed, confidence, judgment, and productivity. But he also acknowledges the danger of emotional dependence and the possibility that AI becomes a substitute for human connection.
Kelly’s core belief
He argues that creativity is not mystical; it is a process. AI, in his view, will:
- enhance creativity
- reveal hidden talents
- help people discover what they are good at
- make creative work faster and more accessible
A Creative Industry Example: Aline van der Velden and Tilly Norwood
The episode then turns to Aline van der Velden, founder of Particle 6, a production company that now works heavily with AI. Her most controversial project is Tilly Norwood, a fully AI-generated “actress” created to be a marketable synthetic star.
Why she created Tilly
Van der Velden argues that AI makes it possible to create new kinds of productions and characters that traditional studios would never greenlight. She sees AI as a tool for:
- cheaper production
- new forms of storytelling
- customized on-screen personalities
- expanding opportunities in underrepresented genres and demographics
How Tilly was designed
Van der Velden describes Tilly as:
- 24 years old
- British
- ethnically ambiguous
- appealing to a broad audience
- smart, witty, and “girl-next-door”
- intentionally not hypersexualized
She frames Tilly as a character built through prompts and iteration, more like a novelist crafting a persona than a studio replacing an actor.
Her argument
Van der Velden insists that:
- AI itself doesn’t create stories on its own
- humans still drive the creative process
- AI is more like a “creative calculator” or enhancer
- the goal is not to replace people, but to create new forms of entertainment
The Academic View: Ethan Mollick on What AI Actually Does
Wharton professor Ethan Mollick offers a more measured framework for understanding AI’s role in work and creativity.
Three phases of AI
He describes the last decade of AI as moving through:
-
Data-driven decision-making
AI used for recommendations, pricing, targeting, and prediction -
Generative AI
Chatbots and systems that can create text, images, music, and code interactively -
Agentic AI
AI systems that can perform extended tasks and complete weeks’ worth of work with minimal supervision
Three modes of human-AI collaboration
Mollick also offers a useful framework:
- Centaur — human and AI split responsibilities cleanly
- Cyborg — human and AI work together in a blended, iterative way
- Conjurer — the human “summons” AI workers to perform specific tasks, then evaluates the results
He says many people start as centaurs and gradually become more blended as they use AI more.
His four “robust human advantages”
Mollick argues humans still have advantages in:
- Deep knowledge — expertise in a field
- Wide knowledge — broad interdisciplinary understanding
- Taste — selecting the best output from many AI-generated options
- Agency — acting decisively in unclear environments
His view: AI won’t just replace people wholesale; it will reshape which human skills matter most.
What the Research Says About Creativity
Mollick and co-authors have studied AI’s effect on creativity and diversity.
Main findings
- AI often scores higher than most humans on creativity tests
- But AI-generated ideas are often less diverse
- AI can become more diverse with better prompting and human guidance
- AI models also show recurring stylistic and conceptual patterns, meaning their creativity is real but constrained
Key insight
AI can generate lots of ideas quickly, but humans are still essential for:
- selecting good ideas
- pushing outputs in unusual directions
- combining AI with lived experience and taste
The Industry Pushback: Sean Astin and Labor Concerns
The episode gives substantial attention to the opposition, especially Sean Astin, president of SAG-AFTRA.
His main concerns
Astin argues that AI is a real threat to performers because:
- entertainment work has already declined sharply
- studios are under pressure to cut costs
- voiceover and digital performance are especially vulnerable
- synthetic performers may erode the livelihoods of human actors
He strongly objects to AI systems trained on what he sees as “ill-gotten data” — material scraped from the creative labor of others without permission or compensation.
His broader argument
Astin sees the issue not just as technological progress, but as a question of:
- labor rights
- consent
- compensation
- whether synthetic “performers” can be treated as equivalent to human artists
Central Tension of the Episode
The episode repeatedly returns to one core question:
Is AI a complement to human creativity, or a substitute for it?
Optimists say
- it will amplify human talent
- it will unlock hidden creativity
- it will help more people make more things
- it may democratize creative production
Skeptics say
- it will replace workers
- it will concentrate power in AI labs and studios
- it will erode originality
- it may make creativity more generic and less human
Takeaways
- AI is no longer just a tool for prediction; it is becoming a creative collaborator.
- The most provocative future version may be the exo-self: a personalized AI that is always present and deeply integrated into your life.
- Creative industries are already using AI heavily, often more than they admit publicly.
- AI can generate impressive creative output, but human judgment, taste, and direction still matter enormously.
- The economic and ethical stakes are high, especially for actors, writers, and other creative workers whose labor can now be simulated.
What to Watch For Next
The episode sets up the second half of the series, where the show plans to:
- interview Tilly Norwood herself
- hear more from economists studying AI and the creative industries
- explore whether the story ends in disruption, adaptation, or something closer to a happy ending