Usage as a moat in AI.
This a slightly more advanced spin on the idea of a compounding advantage from more data. Tomasz Tunguz now offers a method for that advantage to compound: He highlights a recent paper about Reflexion, “an approach that endows an agent […]