How agents will change banking forever | E2260

Today’s show: How long until AI models can improve AI models? Once possible, recursive self-improvement by AI technology could accelerate — forever. Thus far, humans (and their coding agents) are still driving AI progress. But a recent project by AI developer extraordinare Andrej Karpathy, called ‘autoresearcher’, is turning heads as it shows that it is possible — in certain contexts — to allow AI agents to run successive coding experiments to improve specific elements of LLM performance. Call it an early demonstration of the future. OpenClaw is exploding in China, while here in the United …