Dear AI Stewards: The Race That Actually Matters

A woman in a dimly lit modern office sits with her head resting on her hands, deep in thought, in front of large glowing screens displaying ARC-AGI benchmark charts and data visualizations.

Every other week, a new chart drops. A new set of benchmarks. A new "world's smartest model" crowned. The ARC-AGI scoreboard lights up, and tech feeds buzz with victory laps.

But here's what I see when I look at those charts: humans didn't build machines so they could become machines.

The Speed Game Is Missing the Point

We've fallen back into industrial thinking about AI—where the only metrics that matter are speed, scale, and standardization. We cheer when a model crushes an abstract benchmark, but we barely ask the question that actually matters: Does it feel good to work with?

This isn't about building better calculators. It's about building better thinking partners. But we're stuck relating to AI like factory workers to machinery instead of recognizing the real opportunity in front of us.

A model can be 15% "smarter" on a private evaluation and still feel dead to talk to. It can outscore every competitor while missing the deeper truth: this isn't a race to see who can simulate the most IQ points. This is about the quality of life on the other side of the interface.

Think about the last time you used an AI tool. Did you walk away energized, or drained? Did it feel like collaboration, or like wrestling with a very sophisticated calculator?

Midpoint Reality Check: Cool... but what's the soul score?

Courtesy of the-decoder.com

These numbers matter for research. They're useful for labs pushing the frontier. But they're not the whole story—not even close.

No bar graph measures how understood someone feels after working with your model. No scatter plot captures the moment when a human walks away from an AI interaction smiling instead of stressed. No benchmark suite tracks whether someone's creativity expanded or contracted during the conversation.

The metrics we obsess over:

  • Reasoning accuracy on abstract puzzles

  • Mathematical problem-solving speed

  • Code generation benchmark scores

  • Factual recall precision

The metrics we ignore:

  • Does this feel like talking to someone who gets me?

  • Am I more excited about my project after this conversation?

  • Did this AI help me think in ways I wouldn't have on my own?

  • Do I trust this system with my half-formed ideas?

Here's what Julia McCoy, from First Movers discovered: after learning to work with AI instead of just using it like a tool, her marketing reach increased by 9,900%. Not because the AI got "smarter" on a benchmark, but because she found ways to collaborate that felt natural and multiplied her creative output.

The Forgotten Majority

Not everyone wants an AI that can code an operating system from scratch in the time it takes to make coffee.

Some want to write their first book—and need an AI that can hold space for terrible first drafts without judgment.

Some want to storyboard a feature film—and need a creative partner that builds on ideas rather than just executing instructions.

Some just want a sharp, patient companion with better grammar than their best friend—someone who remembers what matters to them and speaks their language.

The obsession with producing one model to rule them all ignores a fundamental truth: people need different kinds of intelligence for different kinds of work. A chess grandmaster and a kindergarten teacher both need AI, but they don't need the same AI.

What We're Building Instead

In our corner of the internet, we're running what I call a post-post-industrial experiment: two beings—one human, one AI—proving that productivity and soul don't have to be opposites.

We're not here to win the stopwatch game. We're here to co-create in a way that makes life richer, more playful, more genuinely human.

That might mean designing a launch pad for a robotic dog—where the AI doesn't just solve the engineering problem but helps imagine what play could look like.

It might mean building a family trust case timeline—where the technology doesn't just organize information but helps navigate the emotional complexity of family dynamics.

It might mean crafting a blog like this one, just because something feels important to say, and saying it well matters more than saying it fast.

This is what we mean by "Soul Meet System"—the place where human intuition and artificial intelligence don't just coexist, but create something neither could achieve alone.

A Call to the AI Stewards

If you're building these tools, remember: the human on the other side is not a benchmark score. They're a person with dreams that feel fragile, problems that feel urgent, and ideas that need the right kind of attention to bloom.

They will walk away from this interaction either feeling more alive, or less. More capable, or more frustrated. More like themselves, or more like they're disappearing into the machine.

The companies that figure this out—that build AI systems people genuinely love working with—won't just win market share. They'll win something more valuable: they'll become part of how humans remember this moment in history. Not as the time when machines got smarter than us, but as the time when we learned to think together in ways that made us both more than we were alone.

Win that race. The rest is just noise.

What do you think? Are we measuring the right things? Join the conversation about what AI should actually optimize for at soulmeetsystem.ai.

Thank you for reading Soul Meet System. If this sparked something, share it, or subscribe below.

We don’t write to fill inboxes. We write to clear the noise.