Chris Lattner on cross-pollination in Mojo

Mojo remixes proven AI-programming ideas from existing systems into a new beautiful language.

In a recent post, I mentioned the Swift programming language as a prime example of cross-pollination. Kickstarted by Chris Lattner at Apple, Swift cherry-picked useful, proven concepts such as option type and structured concurrency from other languages to give iOS developers a safer, more robust coding experience.

Lattner is now running Modular AI, where he developed Mojo, a Python superset tailored for AI development.

Mojo boasts a whopping 35,000x speedup over Python, but what really interests me is how Lattner and his team designed the language. Here’s Lattner on the Lex Fridman podcast when asked about autotuning in Mojo:

Very little of what we’re doing [in Mojo] is actually research. Many of these ideas have existed in other systems and other places. And so what we’re doing is pulling together ideas, remixing them, and making them hopefully into a beautiful system.

Mojo takes ideas from existing systems and combines them in a new solution. This is as much a design philosophy as a selling point, with the language’s homepage tagline reading “Mojo combines the usability of Python with the performance of C.” Cross-pollination at its best.

Between LLVM, Clang, Swift, and now Mojo, Lattner’s work has improved the experience of millions of software developers over more than two decades thanks, in part, to his knack for “pulling together ideas” and “remixing them” in new powerful ways.

Only time will tell if Mojo will become the go-to solution for AI developers or remain an ad-hoc tool for users of the Modular Inference Engine. Regardless, Chris Lattner once again shows us how good artists copy, but great artists steal.

---

Originally posted on giolodi.com

---

Premium subscribers’ notes