Sakana AI, a Tokyo-based R&D company, has unveiled “The AI Scientist,” an AI system designed to fully automate the scientific research process.
According to the company, this system is the first of its kind, capable of independently handling multiple aspects of research, including generating novel ideas, writing code, executing experiments, visualizing results, describing findings in a scientific paper, and running a simulated review process.
“In theory, this process can be repeated to iteratively develop ideas in an open-ended manner, functioning like the human scientific community,” it stated.
This AI Scientist uses advanced Large Language Models (LLMs) to engage with users and suggest new research paths in machine learning. It is open source, governed by the Apache 2.0 license, allowing legal use, modification, and commercialization.
Sakana AI’s research papers cost less than $15 each to produce. The system has a sophisticated simulated peer-review process that evaluates its own work, mimicking traditional peer-review methods.
As a demonstration, The AI Scientist was tasked with completing the entire research process of various established investigations and drafting research papers.
Sakana AI has released several AI models, including an image generation model, a vision language model called EvoVLM, and a model for generating Japanese Ukiyo artwork images.
These models are available on the Hugging Face repository. The company’s primary aim is to develop a novel type of foundational AI model inspired by natural intelligence.
Sakana AI’s announcement of The AI Scientist sparked controversy. Omniscience, another AI startup, argued that Sakana AI was not the first to develop such a system.
They claimed their AI model, Omni, existed before The AI Scientist. Omni assists users by extracting information from their files as they write, aiding in creating content on various subjects.
An example of Omni’s capability was seen in an essay titled “The Unconstitutionality of California Senate Bill 1047,” which was initially drafted using Omni and later reviewed by a human team with counter-arguments.
The team responded to Sakana by providing a copy of their own announcement for The AI Scientist, simply replacing the name with their product. They argued that the same claims could be made about their system, which had been released months earlier.
Stories You May Like