Urgent.News

What's breaking now, across thousands of outlets.

AI

Swedish researchers create an ‘AI scientist’ that designs and runs its own experiments

Swedes created an AI system that can carry out the full scientific process from hypothesis to results interpretation with minimal human input.

Swedish researchers have developed a groundbreaking AI scientist capable of autonomously generating scientific hypotheses, designing experiments, and interpreting results. This AI system, a result of integrating multiple Large Language Models, large databases, and physical laboratory robotics, functions within a closed-loop system to conduct the entire scientific discovery cycle.

According to Ievgeniia Tiukova, a researcher at Chalmers University of Technology and one of the authors of the study, "It is too much information for a human to analyse, but our AI scientist could identify promising biological questions, recommend experiments to test them, evaluate experimental outcomes, and iteratively refine its understanding based on new evidence."

The AI scientist was trained on a database of approximately 60,000 phenotypical, physiological, and metabolic relations for S. cerevisiae, also known as brewer's yeast or baker's yeast. Using this data, the AI generated nearly 2,000 testable predictions regarding how nutrients impact cell growth and stress resistance.

These hypotheses were then evaluated, comparisons were made with control groups, and the AI transformed the ideas into machine-readable instructions for laboratory equipment. Upon completion of the tests, the AI analyzed the data, determined the accuracy of its predictions, and automatically updated or refined its hypotheses for subsequent cycles.

Tiukova emphasized that the AI scientist does not merely serve as a decision-supporting tool; instead, it actively generates new scientific knowledge. The study's senior author, Ross King from the University of Gothenburg, noted that the future of scientific research lies in automation, driven by advances in robotics, AI, and high-throughput technologies.

He believes that AI Scientists will collaborate with human scientists to accelerate discoveries in biology, medicine, and biotechnology, thereby reducing the time required to explore complex scientific questions and optimizing the use of laboratory resources.

While the researchers acknowledge the potential of AI to minimize human bias, accelerate research cycles, and reduce variation caused by human error, they stress that humans will remain essential in scientific research. King highlighted that humans will continue to play crucial roles in setting priorities, interpreting broader contexts, developing research programs, and ensuring compliance with ethical rules.

Future generations of autonomous laboratories will increasingly collaborate with humans and become valuable partners in tackling challenging questions in biology and medicine.

As AI use in research expands, international organizations are establishing guidelines for its responsible and ethical application. The World Health Organization (WHO) recently published a report offering recommendations for researchers, ethics committees, regulators, funders, and policymakers to ensure that AI-enabled health research is conducted responsibly and benefits all.

WHO Director Meg Doherty emphasized that while innovation in AI holds unprecedented opportunities to accelerate health research and improve lives, it must be guided by strong ethical safeguards to protect human dignity, rights, and equity.

Written by urgent.news from Euronews's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at euronews.com →

More in AI

Cloudflare Ships Open-Source Clef Models for Automated Support and Ops Triage

Cloudflare released two open-source "decision models," Clef and Clef-flash, hosted on its Workers AI platform, alongside a new reinforcement learning (RL) fine-tuning product.

  • Cloudflare introduces open-source Clef models for automated support and ops triage.
  • Clef outperforms general LLMs in domain classification speed and category labeling.
  • Cloudflare offers RL fine-tuning service for customizing Clef models with customer data.

LLM Paper Trader

I Built a Small LLM-Powered Paper Trader for a Friend This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend .

  • Small paper trading system created using LLM for financial markets experimentation
  • System combined technical indicators with AI model to make BUY, SELL, or HOLD decisions
  • Open-source Python project on GitHub for accessible experimentation and modification

How I Built a 24/7 AI DevOps Coding Agent for About CAD 200

Hermes + Claude Code + Telegram on a cheap always-on thin client. Hi, I'm Amir 👋 I'm a DevOps engineer, and a surprising amount of my time goes into work that is important but repetitive.

  • Amir built AI DevOps coding agent for ~CAD 200
  • Hermes AI harness delegates tasks to Claude Code
  • System uses Telegram for QA, Git, and reporting

More from Friday 2 October →