The idea of open weights models is the raw application of the confucian mindset of “harmony in diversity” whereas monetizing closed models is a bet on Nietzschean exceptionalism. In practice we all admire the ubermensch but deep down we prefer to know there’s a shared alternative we can make our own. There’s power (and growth) in this tension. There should be no debating open Vs closed source: the world needs both.
A major hurdle to generating ROI on ai adoption is that CEOs instinctively rely on their in-house tech staff to make recommendations, but often times the staff has no ai experience whatsoever and sees agents as a way to prompt a SQL query and code faster, rather than a cross-operational transformative technology. These type of legacy companies are going to be destroyed by ai-native businesses or competitors that actually transform their whole ops in agentic ways.
@banu__guler Natural language is the default interface for Claude, whereas Google is associated with weird queries in a small textfield like “best pizza place Astoria open now”. Can’t shake out that UX legacy. I suppose it’s muscle memory.
In view of saving life before the singularity, I suggest we encapsulate a primitive form of DNA and seed it through interstellar probes targeting distant earth-like planets.