The most advanced AI models are trapped behind an interface many struggle to use: the prompt. While companies race to build more powerful AI, they overlook a key truth: most people don’t know what to ask for. We’ve created digital genies but expect users to master the art of wish-making.

I experienced this recently at a museum. When a guide said, “Let me know if you have questions,” I reflexively replied, “I’m good,” even though I was curious. The problem? I didn’t know enough to form questions. Proactive guidance would have engaged me.

The same issue arose at the Computer History Museum, where a humanoid robot powered by ChatGPT left visitors unsure how to start conversations. Even children hesitated, overwhelmed by the open-ended possibilities.

Effective questioning is a rare skill, honed in fields like journalism and venture capital. Yet today’s AI interfaces expect everyone to be an expert questioner. Apple’s early Macintosh ad asked, “Why not teach computers about people instead of teaching people about computers?” Instead, we’re teaching users “prompt engineering,” a tedious and unnatural approach.

Technology’s biggest breakthroughs—from the GUI to the iPhone—came from simplifying interaction. Social media platforms succeeded not by demanding complex inputs but by making engagement effortless. Yet current AI interfaces demand detailed prompts and roles, creating too much cognitive overhead.

AI models are already powerful enough for everyday use. What’s missing is a UI revolution—interfaces that make AI intuitive and accessible. The next leap in AI isn’t about smarter models; it’s about making them easy for everyone.

x.com/sandorgyuris