In the race for AGI/ASI, product is getting left behind

The race toward Artificial General Intelligence (AGI) and Artificial Super-intelligence (ASI) dominates headlines. But while researchers chase benchmarks, product teams risk leaving users behind.

Right now, AI feels like the command line of the 1980s. To use it effectively, you must learn a specialized language—what we now call a prompt. The syntax may be natural language instead of symbols, but the burden is the same: the user must adapt to the machine.

As a design leader, I’ve seen this pattern before. Powerful new tech often launches in raw form, and it takes design to make it usable at scale.

Some tools hint at the future—Copilot or Claude for coding—but many everyday workflows remain fragile. Take the writing tool in ChatGPT. It doesn’t fail because the models are weak; it fails because it breaks the rules of product design. It overwrites when it should refine, it loses continuity, and worst of all—it makes users fear losing the text they’ve painstakingly written. That’s a cardinal sin of product design: if people don’t trust a tool to preserve their work, they’ll never rely on it.

The result? Clunky, patchwork workflows. Copy, paste, re-prompt, repeat. More juggling windows than seamless assistance.

The real race isn’t just for smarter models; it’s for smarter products.

• Intelligence locked in a benchmark doesn’t change lives.

• Intelligence designed into workflows does.

Eventually, there will be a winner for the smartest model—but in reality, it won’t matter as much as the companies who make those models trustworthy, usable, and invisible in the right moments to human beings.

That’s where design leaders must step in—turning raw capability into human-centered value.