Article illustration 1

In technology and creativity, progress follows two distinct paths: the safe ascent of existing peaks or the dangerous creation of new mountains. This dichotomy—dubbed hill-climbing versus hill-making by Kevin Kelly—frames a critical tension in innovation that resonates deeply across AI development, corporate strategy, and artistic creation.

The Efficiency Trap

Hill-climbing represents the relentless optimization of known processes. It's the domain of:
- Refining algorithms for marginal speed gains
- Perfecting existing product features
- Maximizing ROI on proven technologies

As Kelly notes, this approach "rewards efficiency and optimization" but faces diminishing returns. The peril lies in what economists call the innovator's dilemma: Over-optimized systems become trapped on local maxima, unable to risk descent toward potentially higher peaks. Toyota's mastery of combustion engines, for example, historically complicated its pivot to electric vehicles.

The Art of Mountain Building

Hill-making, conversely, involves creating entirely new domains for exploration:
- Inventing novel architectures (like transformers in AI)
- Pioneering unexplored tech paradigms (quantum computing)
- Establishing new creative mediums (VR storytelling)

"Hill-making entails occupying an area where your work increases the possibilities for more work," Kelly writes. "Inventing the technology of cameras opened up vast realms of visual possibilities."

Article illustration 3

Schematic of adaptive landscapes where hill-climbers optimize peaks while hill-makers create new mountains.

AI's Ascent Limitation

Generative AI currently excels at lower-case creativity—remixing patterns within established domains:
- Generating plausible code variations
- Creating derivative art styles
- Optimizing known solutions

Yet as Kelly observes: "LLMs do not seem to be creating entirely new ways to approach text, images, and tasks." The hill-making deficit stems from AI's inability to:
1. Tolerate efficiency-destroying exploration
2. Pursue curiosity-driven dead ends
3. Strategically "devolve" from local maxima

The Human-Machine Innovation Frontier

The most promising path forward? Combining strengths. Humans provide the hill-making intuition—asking radical questions and embracing productive failure—while AI accelerates hill-climbing execution. This symbiosis could unlock:
- Faster iteration cycles for nascent technologies
- Enhanced tools for scientific discovery
- New creative mediums blending human and machine ingenuity

As Kelly concludes, the future belongs to those who master both modes: optimizing the present while courageously cultivating the unexplored valleys where tomorrow's peaks will rise. For developers and tech leaders, this means designing systems that reward strategic inefficiency—the fertile soil of breakthroughs.