Exin Therapeutics Uses AI Agent For Mouse Brain Surgery
Exin Therapeutics has used an AI agent to perform a brain injection procedure on a live mouse, marking a significant step in integrating AI with laboratory research. This development may lead to more efficient and reliable research methods.
Original reporting · wired.com
Original: “An AI Agent Injected Drugs Into a Live Mouse’s Brain”
Read “An AI Agent Injected Drugs Into a Live Mouse’s Brain” at WIRED AI →
How this was written. Diagnosis Forward wrote this article with a language model from WIRED AI’s reporting (model: @cf/meta/llama-3.3-70b-instruct-fp8-fast); it describes that reporting rather than adding to it. The original, linked above, is the full story.
Key points from WIRED AI’s report
- Exin Therapeutics has successfully used an AI agent to perform a brain injection procedure on a live mouse.
- The experiment aimed to integrate AI with laboratory research and explore the potential of AI in tasks that require direct interaction with equipment.
- The team used two different AI models, Claude Opus 5.5 and Claude Sonnet 5, to see which one would be more suitable for the task.
- The procedure was successful, but it was not without technical issues, including a glitch that disrupted the equipment's coordinate system.
- The team is now testing other models to see which one performs better and plans to experiment with more complex tasks, such as agent-controlled craniotomies.
Biotech company Exin Therapeutics has made a breakthrough in using artificial intelligence in laboratory research. The company's CEO, Gabriel Ocana Santero, recently conducted an experiment where an AI agent was used to perform a brain injection procedure on a living mouse.
The AI agent was given control of a robotic arm and an injector, allowing it to carry out the procedure with some human oversight. This experiment aimed to explore the potential of AI in laboratory research, particularly in tasks that require direct interaction with equipment.
Santero's team used a custom-built setup that included a device to immobilize the mouse and a camera to allow the AI agent to observe and control the procedure. The AI model was trained on published scientific literature, but this experiment showed that integrating AI with real-world experiments can make the models more reliable.
The team tested two different AI models, Claude Opus 5.5 and Claude Sonnet 5, to see which one would be more suitable for the task. However, the Opus model refused to guide the needle into the mouse's brain, citing potential risks to the animal.
The team then switched to the Sonnet model, which agreed to proceed with the procedure. The AI agent selected the treatment, brain region, and injection coordinates, and then controlled the robotic arms and injector needle.
Although the procedure was successful, it was not without technical issues. A glitch disrupted the equipment's coordinate system, and the AI agent refused to modify the code to correct the problem due to its programming restrictions.
The team resolved the issue by using a separate AI agent to make the necessary code changes, but this delayed the procedure. Despite the glitches, Santero and his team were impressed by the AI agent's performance and are now testing other models to see which one performs better. This article is based on reporting by WIRED AI.
Questions and answers
What was the purpose of the experiment?
The purpose of the experiment was to explore the potential of AI in laboratory research, particularly in tasks that require direct interaction with equipment.
Which AI models were used in the experiment?
The team used two different AI models, Claude Opus 5.5 and Claude Sonnet 5, to see which one would be more suitable for the task.
What were the results of the experiment?
The procedure was successful, but it was not without technical issues, including a glitch that disrupted the equipment's coordinate system.
