A new “artificial intelligence scientist” from Sakana AI Labs can automatically create scientific articles, from idea generation to final text, for only $15. Using large language models (LLMs), the system is able to brainstorm, code new algorithms, and write articles without human intervention. But is this good for science?

Automated research, while it can speed up the scientific process, raises concerns about the possible decline in the quality and novelty of work. There is a risk that such systems could flood the scientific literature with papers of dubious value, making it difficult to distinguish between genuine discoveries and “scientific slop.”

Science needs trust and integrity. Are we ready to rely on artificial intelligence in such an important area as scientific research?