AI is often analogized to the brain. But the majority of intelligence exists between our brains, not within them. Language, our greatest ever invention, allowed us to efficiently store, retrieve, and share information across individuals. And every major development in the technology of language—speech, writing, printing, code—has catalyzed a new civilizational era for our species, despite no changes in our brain structure over the past 100,000 years.
A similar progression can be found in the history of computing itself. In the 1970s, computers were capable of incredible feats of calculation—but they were largely isolated from each other. What ended up being the more powerful technology was the internet: the communication protocols that allowed diverse machines to network into ensembles.
We believe the same will be true for AI. The world won't have one single, giant model that everyone sends everything to. Instead, there will be thousands of foundation models, millions of fine-tuned variants of these models, and billions of custom inference engines running on top of them. We believe this is not only a better future for AI, but
a safer one as well.
We started Not Diamond to enable this multi-model future, starting with robust infrastructure for routing between diverse AI models. We’ve been publishing research in the field of AI for more than a decade, and every member of our team has built companies from the ground up before. We’re honored to be backed by some of the world’s greatest AI scientists, founders, and executives. We couldn’t imagine spending our time on anything else.
In modal logic, "not diamond" (or ¬◇ as it's commonly written) translates to “not possible”. At the end of the Tractatus Logico-Philosophicus, Wittgenstein famously writes “Whereof one cannot speak, thereof one must be silent.” We cannot speak about what we cannot speak about. But the limits of expression themselves are not fixed. As the technology of language has evolved, so too has our capacity for expression. And beyond any doubt, the expansion of these limits has produced more value for humanity than any other class of invention or discovery. By networking together intelligent machines, we aim to expand the limits of intelligent expression into what is currently not possible.
We’re a small team based in San Francisco. If you’d like to contribute to what we’re building, we’d
love to hear from you.