Build Smarter Games Without OpenAI

Build Smarter Games Without OpenAI

Former Medal.tv founder Pim De Witte has raised $134 million for General Intuition to explore AI “world models” that could redefine how games are designed, played, and experienced.

Eliza Crichton-Stuart

Eliza Crichton-Stuart

Updated Nov 1, 2025

Build Smarter Games Without OpenAI

When Pim De Witte, founder of General Intuition, walked away from a reported $500 million offer from OpenAI, few expected his next move to be even bolder. Now, with $134 million in fresh funding, In an interview with Deconstructor of Fun, De Witte explains how he sis setting out to transform how video games are built - from predictable and scripted experiences to worlds that think and react on their own.

Redefining Game Design Through World Models

In most games today, everything players encounter - from enemy behavior to dialogue choices - follows a set of predefined rules. De Witte believes that era is coming to an end. The future, he argues, lies in world models, a new generation of AI systems that understand and simulate reality rather than just render it.

While traditional engines like Unity and Unreal rely on deterministic logic, world models introduce an element of unpredictability. They learn how environments respond to player actions by studying massive amounts of gameplay data. Instead of simply displaying what comes next, these models understand why it happens.

This could lead to more immersive and lifelike gameplay. In future titles powered by world models, AI-controlled characters might display human-like hesitation, teamwork, or even panic. Every encounter could play out differently, not because of randomization, but because the system genuinely interprets cause and effect. As De Witte explained, “Determinism makes games stable, but non-determinism creates surprise. That’s what keeps players coming back.”

From Generative to Agentic: A New Phase for Game AI

De Witte sees gaming evolving through three main stages: scripted, generative, and finally, agentic. Scripted systems follow developer instructions. Generative AI creates new assets or environments. But agentic systems - the focus of General Intuition - can act and react based on understanding.

This agentic layer could transform how both developers and players interact with digital worlds. Instead of designing every possible outcome, developers could focus on guiding intelligent systems that adapt in real time. For players, this could mean experiences that feel truly alive - games that learn and respond with nuance rather than rigid rules.

It’s a shift that echoes what’s already happening across creative industries. Just as generative AI is changing how art and music are made, world models could redefine how interactivity itself is designed.

Why General Intuition Stands Apart

Unlike AI research giants such as OpenAI or Anthropic, General Intuition isn’t focused on text generation or chat interfaces. Its mission is to understand worlds - how things move, collide, and change. The company’s foundation comes from De Witte’s previous venture, Medal.tv, a platform that collected over two billion gameplay clips. That dataset now serves as training material for General Intuition’s models, giving the team an unparalleled view into how humans play, react, and strategize.

De Witte also structured General Intuition as a Public Benefit Corporation, emphasizing its goal of supporting human creativity rather than replacing it. “We’re not building AI to take over development,” he said. “We’re building AI that plays alongside developers.” That philosophy resonated with investors, resulting in one of the largest early-stage rounds ever raised in gaming and AI.

The Role of AI-Fluent Studios

A common belief in the industry is that only “AI-native” studios - those built entirely around artificial intelligence - will succeed in the coming years. De Witte disagrees. He believes that traditional developers can compete if they become AI-fluent, integrating new tools without abandoning what already works.

“Good studios already have taste and strong systems,” he said. “AI doesn’t replace that. It enhances it.” This approach could define the next wave of development. Legacy studios that understand how to use AI to speed up iteration, improve agents, and enhance simulation may end up leading the charge - not losing to startups built around AI hype.

Games as Laboratories for Intelligence

De Witte’s broader philosophy extends beyond gaming. He views video games as ideal testing grounds for developing intelligent systems safely and creatively. Games combine perception, reasoning, and action in ways that mirror real-world challenges - without real-world risks.

“Games are where people show intelligence at its highest level,” De Witte said. “That’s why we use them to teach machines how to think.” By treating games as environments for experimentation, General Intuition hopes to guide how AI evolves - toward collaboration rather than control.

The Bigger Picture: AI’s Double-Edged Promise

De Witte’s vision arrives during a time of uncertainty around AI’s economic and social impact. Automation continues to reshape industries, eliminating traditional roles while creating new ones. Companies like Amazon and UPS are introducing large-scale automation programs, while college graduates face increasing underemployment.

Despite the anxiety surrounding these changes, De Witte offers a more pragmatic perspective. He argues that automation is part of a long historical cycle of adaptation. Just as industrial machines replaced manual labor but created new opportunities, intelligent systems could do the same - if society learns to integrate them responsibly.

For him, games offer a blueprint for how that transition might unfold. They are systems built on experimentation, feedback, and human agency. Before AI runs global industries, perhaps it should first learn to play fair in simulated worlds.

The Road Ahead

General Intuition’s next few years will test whether world models can deliver on their promise. If successful, they could reshape not just how games are made, but how intelligence itself is designed and understood.

For now, De Witte’s message is clear: the future of gaming isn’t about replacing developers with algorithms - it’s about expanding what’s possible when human creativity and machine intelligence work together.

Source: Deconstructor of Fun

Frequently Asked Questions (FAQs) 

What is General Intuition?
General Intuition is an AI company founded by Pim De Witte that focuses on building world models for games. These models simulate realistic cause-and-effect behavior, allowing game environments to respond dynamically to player actions.

How much funding has General Intuition raised?
The company recently raised $134 million in seed funding to develop its technology and expand research into agentic game systems.

What are world models in gaming?
World models are AI systems that predict the next state of a virtual world based on player interactions. Instead of simply generating visuals, they understand the logic and consequences behind every action.

How does General Intuition differ from OpenAI?
While OpenAI specializes in language models and text-based AI, General Intuition focuses on spatial-temporal intelligence - understanding how digital environments evolve over time.

Can traditional game studios use AI like this?
Yes. Pim De Witte believes that existing studios can thrive if they adopt AI tools effectively, becoming “AI-fluent” rather than fully AI-native.

Why did Pim De Witte leave OpenAI’s offer?
Reports suggest OpenAI was interested in acquiring De Witte’s previous company, Medal.tv, for around $500 million. Instead, he chose to pursue his vision of building intelligence through gaming.

What impact could world models have on the future of gaming?
If successful, world models could lead to games with more lifelike AI behavior, adaptive storytelling, and environments that feel truly responsive to player choices.

Educational, Reports

Updated:

November 1st 2025

Posted:

November 1st 2025

Leaderboard

View All

Streaming