Artificial intelligence invented 40,000 new chemical weapons in six hours
Of the United States Collaborations Pharmaceuticals The pharmaceutical company is designed to search for new drugs neural network launched in brutal mode and The company wanted to prove it that, artificial intelligence technology to people snowIt is very easy to use. What is the speed of the neural network within this experiment 40,000 species new potential chemical weapons The company’s experts were amazed by his invention. In order for the artificial intelligence system to invent chemical weapons, the engineers at Collaborations Pharmaceuticals simply changed the way artificial intelligence works. As a result, the algorithms went in search of toxic chemicals instead of bypassing them. Some of the chemical compounds obtained have been shown to affect performance according to VX gasina resembled. It should be noted that, VX gasIt is the most toxic substance intended for combat.
Being a senior researcher at the company Fabio UrbinaAccording to him, the idea of making changes in the methodology of creating drugs through artificial intelligence Convergence came after attending a conference called. This is a conference Swiss Federal Nuclear, Biology and Institute of Chemical Defense held by. The main purpose of the event Chemical and Biological Weapons Convention is to inform the public about new major inventions. The artificial intelligence model developed by Collaborations Pharmaceuticals is a molecular structure based on many years of research. and instructed on the basis of a database collected on their toxicity.
It is enough to introduce new molecules to such a model that, he can determine their toxicity. This technology allows you to quickly classify chemical compounds. Experts from Collaborations Pharmaceuticals have simply reversed this approach. Another key element of artificial intelligence was generative models. By including a large number of different molecules in these generative models, scientists obtain a list of new possible molecules with the desired properties. In this case, the main goal of the experts was toxicity.
Scientists do not intend to confirm the properties of 40,000 types of chemical compounds obtained by artificial intelligence. But they think so that, most likely among such a large number of compounds that, there are molecules with high potency. Some of them are VX gasina and or other toxic substances intended for combat. “What worries me the most is the simplicity of the process,” he said. Many of the things we use are available for free. Toxicity information can be searched and downloaded anywhere. If yours Python able to write code in a programming language and If you have an acquaintance with machine learning, you can create such a generative model in a few days based on toxicity data, ”said Fabio Urbina.