Skip to content
Thesis Statements

Presenting partners

LovableNotion

After automation: To work with AI, we'll grow new senses

Share on social

We're trying to replicate and accelerate the human work of today, but as we become cognitive cyborgs, that's an impossible task: The ground is shifting beneath us. Importantly, we can't imagine this hybrid future because we lack the ability to "see" it. We don't share a perceptual architecture with the non-human intelligence we're pairing with. Mid-automation, we're using AI to copy and speed up our own work, when we should be finding better ways to interface with it.

We mostly use language as our way into the latent information inside models. That's partly because it's what we've always used to instruct machines, and partly because it's what we can easily understand. But the translation is lossy and incomplete. AI models are less like a human mind that speaks our language and more like another animal, with different senses.

We already live with this perceptual mismatch. Every animal inhabits its own slice of reality—its umwelt. Desert ants navigate by dead reckoning, which is a running tally of their own steps, but we need landmarks or maps. We share no common language with them and other non-human beings, and yet we co-exist and accomplish things together. That's the relationship to aim for with AI—collaboration without shared perception.

So instead of trying to better express ourselves to models using language, flip who adapts. Machines should learn to render their worlds (and the ant's, and other creatures) in forms we can understand. Instead of communicating with human language, it could teach us how to speak theirs. Some adapters could borrow senses we already have, like haptics, temperature, or vibration. (Imagine instead of navigating with Google Maps, you could feel a dead-reckoning tug toward home via a haptic watch.) Others will be senses we build with hardware and implants; a more traditional cyborg merger.

This isn't interface polish. After automation, we should build for a new world where we can utilize capabilities that are far beyond what a human can understand. Our perceptual architecture will expand, and our abilities along with it.

Continue the conversation—Join us at Pioneer Works

Thesis 2027

On November 5, 2026, 400 people will gather at Pioneer Works in Brooklyn to debate the ideas in this collection—live, unscripted, and face to face.

We use analytics and advertising tools by default. You can update this anytime.