fuzzy microprocessors

Intel released their first microprocessor in 1971. The last 50 years of computing—the PC, Internet, cloud, and mobile eras—were built on a foundation of precise instructions and strict syntax.

LLMs are something different. They're fuzzy microprocessors that can understand our native languages and operate on human instructions, intentions, or ideas.

questions

How should we interact with these new fuzzy computers? How do we program them? What new primitives do we need? What patterns will emerge?

What if we took some of computing's greatest hits—compilers, REPLs, cron jobs, MapReduce, React—and made them fuzzy? What would they look like?

We've spent the last decade building powerful black box algorithms that few people understand. What if we instead built systems that could be programmed in plain English? Could more people participate in shaping the technology that shapes us?

what's next?

I'll be experimenting in public and writing as I go. If you're curious, you can follow along here.