Until recently, AI’s role in research felt like having a useful assistant. It could summarise a paper, clean up a dataset or draft an abstract. Researchers were still in charge of the thinking.
That changed in late 2025 when cutting-edge “frontier” AI models became capable of reasoning and planning reliably by themselves. A key feature of these models is “tool calling” – the ability to interact with external tools in order to act on the world, not just describe it.
This marks the rise of agentic AI: systems that do not just respond to instructions but can independently plan, execute and iterate. In science as in other fields, chatbots have become coworkers that can autonomously complete real work, end to end.
An example of this is Tokyo-based Sakana AI’s The AI Scientist. Unveiled in mid-2025 and now in its second iteration, the Japanese tech company bills this as “the first comprehensive system for fully automatic scientific discovery”.
The AI Scientist scans existing literature, generates hypotheses, writes and executes code, analyses results and produces a full research paper – largely without human involvement. It reasons, fails and revises, just as a junior scientist would.
