Richard Cornelius Suwandi

When we raise the temperature, we make an LLM agent’s outputs more diverse, but not necessarily more exploratory.

Bayesian optimization (BO) is the standard method for sample-efficient search of expensive, derivative-free black boxes.

Most AI systems are judged by a simple test: given observations they have not seen before, can they predict the right answer?

In my previous post on the Darwin-Gödel Machine (DGM), we have explored how an AI agent can iteratively rewrite its own code, evolving better tools and strategies without humans redesigning it from scratch.

A young girl sits, not in front of a screen, but within a world of her own making.

One of the most thought-provoking moments for me at ICML 2025 didn’t come from a new architecture or a scaling law.

Why do humans keep inventing new musical genres, designing cities that float on water, and imagining life beyond our universe?

Imagine if we could build an AI that thinks and plans like a human.

Most AI systems today are stuck in a “cage” designed by humans.

Large language models (LLMs) have rapidly become indispensable AI assistants.