Can artificial intelligence replace animal testing in the lab?

Researchers are working on an AI application that can analyze the results of animal tests available worldwide and thus avoid unnecessary new tests. These technologies offer hope for a transition to more ethical methods.

No one likes to see animals undergo scientific testing, from animal lovers to lab technicians. Yet ensuring the safety of drugs and other substances for future human use has long justified these practices. However, researchers have been working on non-animal alternatives for decades, and now artificial intelligence (AI) systems are accelerating this work. A simple but promising application of AI is to analyze the results of existing and globally available animal tests. This method aims to avoid the need for unnecessary new tests. Scientists can struggle to sort through decades of data to find and analyze exactly what they are looking for. AI, like the ChatGPT model, could extract and synthesize all this data, thus optimizing its use.

Thomas Hartung, a professor of toxicology at Johns Hopkins University in Baltimore and director of the Center for Alternatives to Animal Testing, argues that AI is as good, if not better, than humans at extracting information from scientific papers. According to him, the need to test new chemicals, with more than 1,000 new compounds entering the market each year, justifies the importance of AI in this field. Trained AI systems are even beginning to determine the toxicity of new chemicals, allowing for rapid preliminary assessments.

Testing human drugs on animals can sometimes be of little use or even dangerous. For example, an arthritis drug passed animal testing before being pulled from the market because of increased risks of heart attacks in humans. Conversely, drugs like aspirin have reportedly failed animal testing despite their common use in humans.

In some cases, AI is already proving to be more accurate than animal testing. Projects like AnimalGAN, developed by the US Food and Drug Administration, aim to replace animal testing by using AI to predict rats’ reactions to chemicals. Another project, Virtual Second Species, creates a virtual dog trained with historical testing data on real dogs. In short, while new technologies like AI cannot yet completely end testing, they offer hope for a transition to more ethical methods.