Sakana AI Labs announced the creation of an AI Scientist last week, an artificial intelligence system that can make scientific discoveries around machine learning in an automated way.
The AI Scientist uses generative large language models (LLMs) such as ChatGPT and other AI chatbots and comes up with ideas as well as creating new code algorithms, plot results, and writes down a summarized paper of an experiment and the findings with reference.
It is known that scientific discovery is one of the most human activities because scientists first need to understand the knowledge and find out a crucial gap. After that they must create research questions, design, and conduct an experiment to find an answer.
They must then analyze and interpret the results of the experiment which could possibly lead to another research question. The AI tool can manage the lifecycle of a scientific experiment at just US$15 per paper.
Scientific research is conducted in the open and nearly all information is recorded in writing to make it accessible and verified. There are countless scientific papers that can be accessed for free on platforms like arXiv and PubMed
LLMs that have been trained using this data are able to understand and replicate the unique language and patterns found in the field of science. Hence why it is not unexpected that a generative LLM can generate content that is of high-quality research paper, given its access to multiple examples that it can copy.
All these AI tools aim to help scientists do their job more efficiently and not to replace their jobs or them.
Main Image: The Conversation
