AI research leads to AI that can conduct research independently. That’s the dream of Japanese company Sakana AI, which claims to have shared the first “AI Scientist.” Designed to conduct autonomous scientific research, its embedded generative AI model could lead to discoveries and innovations faster than human researchers alone.
Sakana collaborated with the University of Oxford and the University of British Columbia to design the AI Scientist and published a research paper paper about what they’ve done so far. The AI scientist would be able to emulate the entire scientific discovery process, including everything from coming up with ideas to designing and conducting experiments and even writing full scientific papers describing the findings for publication. Sakana said it would also be a steal, as the model can produce a complete research paper for as little as $15.
All it needs is a broad topic and access to data to get started. The AI can then look at previous research and look for unanswered questions and ways to produce experiments based on those ideas. It can then loop around and refine its ideas based on the outcomes of its previous research.
For now, the AI Scientist is focused on machine learning, an appropriate choice since that’s the seed of his own creation. So far, the AI Scientist has come up with previously unexplored techniques for different language models and ideas for analyzing how a model evolves through training. That could lead to better machine learning algorithms and, in a sense, the AI Scientist’s successors.
“The AI Scientist automates the entire research cycle, from generating new research ideas, writing all the necessary code, and running experiments, to summarizing experimental results, visualizing them, and presenting the findings in a full scientific manuscript. Sakana explained in a blog post after.
“We believe this work marks the beginning of a new era in scientific discovery: bringing the transformative benefits of AI agents to the entire research process, including that of AI itself. The AI Scientist brings us closer to a world where endless affordable creativity and innovation can be unleashed on the world’s most challenging problems.”
Science AI
The AI scientist could theoretically expand into almost any field of research. Environmental protection, medical research, materials science, and more could benefit from the model’s efforts. However, such a revolution in science is still a long way off. AI software is about the limit of AI’s expertise, and expanding its capabilities into other scientific disciplines would require much more work. As with any other AI model, hallucinations could occur, meaning the research would be incorrect or downright nonsensical and worthless for scientific research.
Sakana isn’t the only one who sees how AI could one day do its own research. In fact, there could be a lot of talk about it soon. The frenzy of rumors surrounding the upcoming OpenAI project “Strawberry” suggests that the ChatGPT creator will have his own AI models that can do research on their own, too. Granted, the idea for Strawberry seems more about the AI doing follow-up research online and thinking ahead to anticipate human questions, but it’s easy to see how that could be adapted to scientific research.