The AI development community is grappling with fundamental questions about how intelligence emerges and how we should build with these new capabilities.

RETHINKING INTELLIGENCE BEYOND LANGUAGE

Yann LeCun made a striking argument today about the nature of intelligence. He pushed back against the language-centric view that dominates current AI development.

LeCun emphasized that true intelligence relies on multiple sensory inputs—touch, hearing, smell, taste, and proprioception. These aren't just nice-to-haves. As he put it, you can live without sight and hearing, but not without the others.

Here's what's particularly interesting. LeCun pointed out that animals develop intelligence without language—almost all of them do. Even humans become smart without nearly as much language exposure as large language models receive during training.

This challenges the assumption that scaling language models is the primary path to artificial general intelligence. It suggests we might be missing crucial components by focusing so heavily on text.

AI-POWERED PROTOTYPING TRANSFORMS DESIGN

Meanwhile, Simon Willison highlighted how AI is fundamentally changing the design process itself. He shared insights from Jenny Wen at Anthropic, who argues that traditional design workflows—user research leading to personas leading to wireframes—may be outdated.

The core hypothesis is provocative. In a world where anyone can make anything, what matters most is your ability to choose and curate what you make.

Willison connected this to programming, noting that AI-assisted development dramatically reduces the cost of building the wrong thing. Previously, a design mistake could waste months of development time. Now, if a wrong direction only wastes days, teams can take more risks and explore problems more actively.

But there's a deeper insight here about problem recognition. As one observer noted, most people's problems aren't software-shaped. Programmers are trained to see automation opportunities everywhere—renaming files, processing data, scripting repetitive tasks. The rest of us remain blind to solutions we were never taught to see.

This suggests that AI's biggest impact might not be making programming easier for programmers. It might be making programming accessible to people who never thought their problems could be solved with code.

These perspectives reveal a tension at the heart of AI development. Are we building toward human-like intelligence that integrates multiple senses and embodied experience? Or are we creating powerful tools that augment human problem-solving in entirely new ways? Perhaps the answer is both—and that's exactly what makes this moment so fascinating.