AI Isn't Stalling, We Just Need to Change Our Perspective
New LLM models keep launching with massive investments, yet they no longer feel magical. Perhaps the stall isn't in the technology, but in how we define 'progress'.
Reading about AI today, an interesting thought came to mind. New LLM models—ChatGPT-5, Grok-4—keep launching with massive investments and sky-high expectations, yet they no longer seem to carry that magical feeling they once did. Growth charts are starting to flatten. People are beginning to talk about a "plateau."
I wonder: what if the plateau isn't in the technology itself, but in how we define "progress"?
The Tree Metaphor
We're like gardeners obsessed with a single metric: the height of the tree. We pour all our effort into making it grow taller, faster, reaching for the sky. But a truly healthy tree doesn't just need height. It needs roots deep enough to grip the earth and a canopy wide enough to commune with sunlight—to provide shade for the entire garden.
Reaching Higher — Reasoning Ability
"Reaching higher" represents the leap in reasoning capability. As models expand to tens of billions of parameters, their chains of thought grow thicker: they help us with everything from daily tasks to problems that were once exclusively human domain.
But reasoning is only half the picture. The thickness of thought cannot substitute for understanding the world—context, causality, interwoven relationships. A tree can reach high, but it's the deep roots that help it stand firm through all seasons.
The Roots — Deep Understanding
The roots represent the capacity for enduring understanding of the world. Today's AI models can write poetry, code, and solve complex problems, but how deeply do they truly "understand" the world they process?
They learn from massive digitized datasets, but how do they reach the deep veins of knowledge—philosophy, art, lived experiences accumulated over millennia? Humanity's data falls like rain: some soaks into the topsoil, some evaporates or flows away; only a portion settles into underground springs.
To reach those springs, roots must go deep enough.
The Canopy — Broad Coverage
The canopy represents AI's ability to cover and apply broadly. A tree with a wide canopy can provide shade for the entire garden, rather than focusing on a single towering trunk that only a skilled few can climb.
In developing AI, instead of building massive models for a single domain like coding or mathematics, perhaps we need smaller, more specialized models that can spread across many different fields—healthcare, logistics, agriculture—and benefit more people.
The Pause Moment
In any developing system—whether technology or nature—sustainable growth is never a straight line upward. It always includes periods of "pause," necessary stillness to consolidate foundations and expand influence.
Perhaps the AI industry also needs such a moment of "pause." A moment to stop obsessing about building more floors higher, and instead ask: how do we dig deeper? How do we spread wider?
AI isn't stalling at all. It's just waiting for us—its creators—to stop looking up at the sky and start looking within. And that's when AI truly becomes part of humanity's garden of life.