A new study observed an artificial intelligence system that created 40,000 chemical weapons that included various compounds in just six consecutive hours following a set of commands from the experts.

AI Solutions Gone Wrong

Chemical weapon by AI
(Photo: imustbedead from Pexels)

The latest AI study presented a power that a super-intelligent program possesses. In a recent paper, experts witnessed how an artificial intelligence analyzed potential chemical compounds that would be helpful to cure global diseases.

In a surprising turn, the particular artificial intelligence developed solutions not just through medicinal aspects but a lot of alternatives based on chemical weapons.

The study demonstrated how modern AI technology could possibly cause negative implications not just for isolated populations but for the whole world, even when initiated by human controls in an unintentional approach. Specialists led the research from North Carolina startup Collaborations Pharmaceuticals.

The aim of the examination, according to Daily Mail, was to determine how an AI algorithm could easily divert its potential solutions from one track to another, suggesting answers that are far more lethal than what people expect from it.

The team focused on possible ways that could abuse an AI-powered program and its corresponding effects when pushed to work on negative tasks by human influence rather than positive ones.

The AI study discovered that an extremely intelligent algorithm could develop thousands of new chemical combinations when taken advantage of through absurd commands. The Verge reported that most of the chemicals assembled by artificial intelligence were similar to the most dangerous chemical weapons still in theory today.

ALSO READ: Clearview AI Face Recognition That Stores People's Images at Work in Ukraine Against Russian Forces


Lethal Chemical Weapons Result of Simulated AI Abuse

Some of the chemical compounds invented by the AI system were pretty close to the properties of venomous agent X or VX, which is known as a toxic nerve agent that initiates the human body to twitch via small doses.

According to the authors, the discovery is truly terrifying, considering how quickly and easily the AI consolidated information from chemicals available today and curated a dataset that included toxic chemicals. These chemicals are dangerous and not a laughing matter, having destructive capacities similar to biological weapon formulas we know and still theorize today.

Collaborations Pharmaceuticals expert and lead author of the study, Fabio Urbina, explained that their team has gone through a lot of molecule datasets to check if they are toxic or not. The VX, which is an acetylcholinesterase inhibitor, caught their attention.

Acetylcholinesterase is utilized by neurons whenever there are activities related to the muscles of the body. It is responsible for telling the brain how to move the chunks wrapped around the entire physiology of humans.

When VX is administered to a person, their lung muscle, also known as the diaphragm, will stop moving, and this will start the organ to become paralyzed and void its functions.

The idea of slipping the switch of AI to cause negative outcomes is presented at the Convergence Conference, an international council organized by the Swiss Federal Institute for Nuclear, Biological, and Chemical Protection.

The study was published in Nature Machine Intelligence, titled "Dual use of artificial-intelligence-powered drug discovery."

RELATED ARTICLE: Microsoft's PeopleLens Use AI to Help Blind Kids Improve Social Interaction

Check out more news and information on Artificial Intelligence in Science Times.