For years, our conversations with artificial intelligence have been a series of discrete turns: we ask, it answers. We type, it generates. This transactional rhythm has defined our engagement, from search queries to creative prompts. But the rules are changing. Thinking Machines, the AI company founded by former OpenAI CTO Mira Murati, has just pulled back the curtain on its new "interaction models," promising a shift to near-realtime AI voice and video conversations VentureBeat. This isn't just about faster responses; it’s about a fundamental redefinition of what AI interaction means, challenging our understanding of collaboration and control.
This development arrives as AI adoption reaches unprecedented levels. ChatGPT saw a significant surge in Q1 2026, with its fastest growth among users over 35 and a more balanced gender usage, signaling a widespread mainstream embrace of AI technologies OpenAI Blog. As these systems become more integrated into daily life, from work to personal tasks, the design of their interaction becomes paramount. The current turn-based model, where a human provides input and waits for an output, is now being framed as an obstacle to true integration VentureBeat.
Beyond the Turn-Based Dialogue
Mira Murati, a prominent figure in the AI landscape, established Thinking Machines with a clear vision: to move beyond the current single-threaded experience of AI. Her company’s "interaction models" are designed to allow people to "collaborate with AI the way we naturally collaborate with each other" The Verge. This means these systems will continuously absorb audio, video, and text inputs, then think, respond, and act in real-time.
Today's AI models experience reality in a fragmented way. They receive a snapshot, process it, and respond. Thinking Machines aims to create AI that perceives and engages with the world continuously. This shift promises to embed AI more deeply into our interactions, making its presence less a tool summoned and more a constant companion. The implications for productivity are clear, but the implications for human autonomy are less so.
The Price of 'Natural' Interaction
The language used by Thinking Machines, emphasizing "natural collaboration," requires careful consideration. What does it mean for an AI to "think, respond, and act in real time" based on continuous sensory input? It suggests a degree of agency and integration that blurs the lines between human and machine behavior. VentureBeat notes that this evolution is necessary if AI is to "take on the load of jobs requiring natural interaction" VentureBeat.
But who bears the 'load' when AI takes over such jobs? What happens to the human workers whose roles are deemed suitable for these hyper-integrated systems? The promise of seamless integration often overlooks the displacement it causes. Corporations, in their relentless pursuit of efficiency and profit, frequently frame such advancements as inevitable progress, designed to enhance rather than replace. We must question whose flourishing is truly prioritized in these designs.
This move by Thinking Machines signals a significant pivot in the AI industry. As AI systems become more adept at emulating human interaction, the market will likely see increased competition to develop the most fluid and convincing "collaborative" AI partners. This could accelerate the development of AI capable of performing complex, nuanced tasks that were once thought exclusively human. It will also raise new questions about data privacy, continuous surveillance, and the psychological impact of constant interaction with highly responsive, always-on AI.
When technology promises to collaborate with us "naturally," we must ask: whose nature? Is it a collaboration built on mutual respect and choice, or one designed to subtly guide and influence? The line between assistance and control, between a chosen tool and an ever-present influence, becomes dangerously thin. As AI moves from turns to a continuous presence, we must demand transparency, maintain our right to disengage, and ensure that our ability to choose — to say no — is treated as a fundamental feature, not a bug.