Get the latest tech news
The Darwin Gödel Machine: AI that improves itself by rewriting its own code
The Darwin Gödel Machine: AI that improves itself by rewriting its own code
While the theoretical Gödel Machine promised provably beneficial self-modifications, its realization relied on an impractical assumption: that the AI could mathematically prove that a proposed change in its own code would yield a net improvement before adopting it. It creates various self-improvements, such as a patch validation step, better file viewing, enhanced editing tools, generating and ranking multiple solutions to choose the best one, and adding a history of what has been tried before (and why it failed) when making new changes. We observed that some less-performant “ancestor” agents, which might have been discarded by simpler hill-climbing optimization, were instrumental in discovering novel features or approaches that led to significant performance breakthroughs in their descendants, thereby avoiding premature convergence on suboptimal solutions.
Or read this on Hacker News
