San Francisco-based startup Mirendil has raised $200 million without having a product or revenue. Its goal is to develop artificial intelligence capable of automating AI research, allowing small universities and laboratories to create specialized models without large teams. For the public, this could accelerate advances in medicine or materials, although the risk is evident.
The technical challenge of an AI that researches 🧪
The project aims for the AI not only to process data, but to formulate hypotheses, design experiments, and draw conclusions. This involves mastering areas such as reinforcement learning and the generation of new knowledge. However, the lack of a minimum viable product and the absence of revenue raise doubts. The promise is that small laboratories can compete with tech giants, but the technical path is uncertain and requires constant validation.
$200 million for an AI that still doesn't know how to turn itself on 🤖
Mirendil has achieved something many startups envy: raising a fortune without having shown anything tangible. Its plan is for the AI to research on its own, but for now it only researches how to spend the money. If the project fails, at least it will have demonstrated that asking for funding is easier than programming. Meanwhile, universities sit and wait for the machine to decide to get to work.