Nye's Digital Lab is a weekly scribble about creativity at the intersection of AI & Distributed Networks.

This week I'm paying attention to that "gut reaction" when I see robots walking around, and thinking maybe we should be considering what it means.

I took my kids to a mediocre Italian restaurant recently. One of those "dad-takes-kids-to-dinner outings" that's more about the ritual than the food. My oldest, making conversation, asks me about... AI.

Instead of explaining the whole realm of AI to my kids, I tried to make it concrete:

"What if our waitress was a robot?"

The horror on his face was immediate.

"That's not good at all," he said. "Why would we even do that?"

Kids cut through BS. There is something fundamentally wrong about humanoid service robots, once you sit with the idea. It's not just job displacement (though Amazon's warehouse automation might replace 160,000 positions). It's what we lose when we swap human interaction for simulated human interaction. It's the creeping sense that we're building a world where genuine connection becomes increasingly optional, mediated, or replaced by something that looks right but rings completely hollow.

My kid's horror has stayed with me because the robotics sector is back at it. After six months of Boston Dynamics' dog-like robots doing backflips and navigating disaster zones, my timeline has pivoted hard toward humanoid robots. Tesla, Figure AI, Agility Robotics, and a dozen startups you've never heard of are racing to build machines that walk on two legs, have faces, and increasingly, personalities.

What's making this humanoid fantasy technically possible? The Jetson Nano and its beefier siblings. NVIDIA's compact AI modules can now fit enough intelligence into a robot skull to run sophisticated neural networks in real-time. And they're affordable.

The barrier to entry has dropped from "well-funded research lab" to "enthusiastic grad student with a credit card." The robots are definitely coming. It looks like they will walk here. But here's what keeps nagging at me:

why are we so hellbent on making them look like us?

Why invest so heavily in creating machines in our own image when tank treads and multiple arms would be far more efficient? I think the answer says more about our loneliness and cultural fantasies than about functional design.

That creepy feeling my kid gets. That's not a bug, it's a feature. Call it a human "gut check" or evolutionary intuition, but our discomfort with humanoid robots exists for a reason. Maybe we should listen to it before we've mass-produced 100,000 uncanny valley inhabitants and scattered them across every warehouse, hospital, and suburban home in America.

Just sayin'...

If you asked a guy who watched Battlestar Galactica whether robots should look human, they might immediately think of Number Six.

Right, that one. The Cylon in the red dress. And yeah, of course part of the population wants our robots to look like that. We’re human. We’re wired to respond to human faces, human proportions, human um… attractiveness.

But isn’t that exactly the problem?

The appeal of humanoid robots isn’t really about functionality. It’s because she’s hot. That's not a functional robot, that's a fantasy robot.

That's the horror of the Westworld promise: machines that look and act human enough that we can project whatever we want onto them, use them however we want, and never have to deal with the messy reality of actual human relationships. There’s something subtly ... overtly ... dystopian about that impulse, even if we dress it up in the language of innovation and user experience.

The human form factor isn’t the most efficient design for most tasks.

You want a warehouse robot? Tank treads and multiple arms beat two legs every time. You want a disaster response robot? The quadruped design (that had its moment before humanoids took over) actually makes way more sense for unstable terrain. The push for humanoid robots isn’t driven by engineering efficiency, it’s driven by our need to see ourselves reflected back, to create something we can relate to.

Because suddenly every robotics startup has access to the computational horsepower that used to require server racks. The hardware constraints are gone. The cost barriers are dropping. Which means the only thing standing between us and a world filled with humanoid robots is...

us deciding whether that's actually a good idea.

My kids talk to “Ziggy” (their renamed Alexa) every day. For weather and Dodgers scores. I'm fine with that. That’s utility.

But the moment my kids start telling Ziggy about their bad days at school, or developing what feels like friendship with an algorithm, we’ve crossed into territory that should make us uncomfortable.

There’s a generation of kids right now who are growing up having their first extended conversations not with humans, but with AI chatbots. As a technologist, but mostly a dad, this is my absolute nightmare.

There are documented cases of teenagers forming emotional dependencies on character AI platforms, treating these bots like therapists, best friends, or what's really bad... their romantic partners. We’re normalizing the idea that meaningful relationships can exist with things that fundamentally cannot care about us.

When we build humanoid robots with personalities, we’re taking this problem and making it physical. We’re creating a presence in the room that acts human enough to trigger all our social instincts.

We’re social creatures.

We’re programmed to read faces, interpret body language, form attachments. A sufficiently "human-like robot" will hijack these systems whether we want it to or not.

Rosie from The Jetsons wasn’t attractive. She was basically a talking trash can on wheels with arms. And you know what? She cleaned the house just fine. The show even lampshaded how weird it would be to have a personality-driven relationship with your household appliances. Yet here we are, decades later, racing toward exactly that future.

When you put a personality on something human made, it should be clear it is designed for entertainment or amusement. Mixing functionality and personality, are two domains that probably shouldn't cross.

We filled the world with cars, then airplanes, and we’ll fill it with robots too.

That’s how technology works. Once something becomes possible and profitable, it becomes inevitable. Some version of these machines will be in warehouses, hospitals, and will be in our homes at some point. Who knows when that is. At this rate, maybe next week?

1X Technologies is targeting thousands of home deployments of their NEO robot by the end of 2025, pricing it "about the same as a modest car."

We need to draw some lines now, while we still can.

Robots shouldn’t look like us. Not really. They should be obviously mechanical and functional, maybe even aesthetically pleasing in their own way, but not human. The uncanny valley exists for a reason. We should stay on the far side of it deliberately, not as a limitation to overcome but as a boundary to respect. Make robots that are clearly tools, not potential companions.

And robots shouldn’t act like us either. They shouldn’t have personalities designed to form emotional bonds. They shouldn’t crack jokes to make you laugh, or offer sympathy when you’re sad, or do any of the thousand small things that make human relationships meaningful. These things can be programmed, but they can’t be meant.

And a relationship with something that can’t mean anything is just elaborate loneliness with better interface design.

We need to preserve robots as tools for as long as humanly possible. The moment we start assigning them real social value, we’ve opened a door to chaos we can’t predict. Do they get rights? Do we owe them consideration? Can you be cruel to something that acts hurt but can’t actually suffer? These aren’t just philosophical puzzles. They’re questions that will reshape society in ways we can’t anticipate.

Can we leave all the personification and just make robots look like robots?

The most functional robots won’t look like us. They’ll look like what they need to be to do their jobs well. And the healthiest society won’t be the one with the most convincing artificial humans, I hope it’ll be the one that remembered to keep the line between tool and person bright and clear. We can have useful robots without humanizing them. We should at least try.

Because once we’ve filled the world with machines that look like people, act like people, but aren’t people?

There’s no going back from that.

That's it for this time. I do this every week. If you vibe to the ideas I express, consider subscribing or sharing with friends. Thanks for reading.

Nye Warburton is an educator from Savannah Georgia who wants it on record that he, for one, welcomed the robot overlords. This essay was improvised using Otter.ai and refined and edited with Claude Sonnet 4.5.

For more information visit: https://nyewarburton.com

Key Companies and References:

• Agility Robotics - Digit humanoid robot

• Figure AI - Figure 02 robot working with BMW

• Boston Dynamics - Atlas and Spot robots

• Tesla - Optimus humanoid robot

• Apptronik - Apollo robot for industrial use

• 1X Technologies - NEO home robot

• NVIDIA Jetson Platform - Edge AI computing modules

• Unitree Robotics - Affordable G1 humanoid ($16K)

• UBTECH Robotics - Walker series and Una robots

• Engineered Arts - Ameca hyper-realistic robot

• My Robot Dog and Me, January 28, 2025

• The Revolution will be Automated, September 18, 2024

• Can AI teach my students? September 14, 2025