John Cutler Draws Parallels Between AI and Improv to Explain Complex Dynamics

In a recent LinkedIn post, John Cutler explores the intricate relationship between Artificial Intelligence (AI) and the art of improvisational theater, drawing fascinating parallels to explain the underlying principles that make both feel remarkably real. Cutler, a prominent voice in the tech and product development space, uses improv as a lens to demystify the perceived “magic” of AI.

Cutler begins by highlighting a fundamental rule of improv: “build, don’t deny.” He illustrates this with a simple example: when faced with a crisis like a sinking ship, an improviser wouldn’t deny the reality but would instead build upon it. As Cutler explains:

“Yes… and the escape pods are gone.”

This principle, Cutler argues, is crucial for creating believable interactions, whether on stage or in conversation. He notes that in real-life dialogues, people tend to build on each other’s contributions rather than immediately invalidating them, fostering a “continuity of reality” that makes the exchange feel authentic.

The Power of Reaction Over Planning

A key aspect of improv that Cutler connects to AI’s emergent properties is the emphasis on reaction rather than rigid planning. He points out that skilled improvisers are trained to listen intently and react honestly, allowing the scene to unfold organically. This contrasts sharply with trying to predict or control every outcome.

According to Cutler, this spontaneous unfolding contributes to the sense of realism because:

• No one is jumping ahead to an outcome
• The moment unfolds based on what just happened

He further elaborates on the commitment required in improv, stating that performers commit 100% to the given situation, even if it’s absurd. Cutler provides a humorous example:

“You’re a banana.”

You respond:
“Wait… I can’t feel my arms… why can’t I feel my arms??”

This willingness to fully inhabit a scenario, however strange, is vital for building a shared reality.

Establishing and Protecting Shared Context

Cutler emphasizes that effective improv groups excel at establishing and protecting shared context. They constantly answer implicit questions about their environment, relationships, and immediate priorities. However, this is often achieved through actions and interactions rather than explicit verbal explanations.

As Cutler notes, this process:

• Builds a shared mental model in real time, together.

This collaborative construction of reality is a core element that makes improv compelling.

Constraints, Coherence, and Collaboration

Contrary to what some might assume, improv is highly constrained, which paradoxically leads to greater coherence. Cutler explains that improvisers must:

• Stay consistent with what’s been established
• Don’t introduce random contradictions
• Heighten what’s already there

These constraints help create a believable world, even when it’s invented on the spot. Furthermore, the principle of making each other look good is paramount. Cutler states, “Support your scene partner.” This involves not stealing focus or setting traps, but rather giving partners opportunities to shine.

“So instead of trying to win, they’re trying to build the scene together.”

This collaborative spirit, combined with subtle coordination like mirroring tone and energy, and picking up on micro-signals, creates a dynamic that feels both spontaneous and structured. Cutler likens this to “social metacognition in action.” By heightening existing patterns rather than restarting, improv creates a sense of inevitability and structure.

AI vs. Improv: Simulation vs. Reality

Ultimately, Cutler posits that improv feels “real” because it possesses all the ingredients of genuine human interaction. It demonstrates how simple rules can combine to create a seemingly scripted experience, but with the crucial difference that participants are actively sharing and updating a real situation.

He concludes by drawing a direct comparison to AI, suggesting that Large Language Models (LLMs) simulate this effect algorithmically and statistically, rather than through genuine shared experience. As John Cutler articulates in his post, the “magic” of AI lies in its ability to mimic the emergent, collaborative, and context-driven nature of human interaction, much like the art of improv.