I believe that the world is entering a very uncertain time in the history of humanity. This concern is driven by the development of AI. An article in Scientific American outlined my fear in a nutshell.
A couple of programmers at Collaborations Pharmaceuticals built an AI application to look for new drugs (a.k.a. molecular structures) that could cure rare diseases. It worked! The system outputs many molecular structures that could bring promising medicines to uncommon, deadly illnesses – especially those so rare that pharma companies would not invest in the research.
Then the programmers added one more filter to AI: remove all molecular structures with dangerous side effects when trying to cure the disease. Proudly, they published their remarkable accomplishments that could reveal hope to so many people suffering from the incurable.
That publication caught the eye of some folks running a conference on biological and chemical weapons. At first, the AI developers were puzzled by this invitation until the lightbulb went off. If their AI could find molecules that cure diseases, it could just as easily create a list of molecules that would make diseases – biological weapons.
The programmers asked their AI-based MegaSyn software to do just that. Provide a list of biological weapons similar to VX, a notorious nerve agent. Within six hours, they had a list of 40,000 substances. VX was used to assassinate the half-brother of the Korean ruler, Kim Jong Un. A VX nerve agent that is so potent that one or two salt-sized molecules can kill in less than an hour. As most of us know, the difference between medicine that cures and medicine that kills is dosage.
ChatGPT is one year old. This one AI tool alone is already reshaping our world.
Think of what could be next, as AI will reshape the world in ways we have not imagined – and it will do so very quickly.
While most people will use AI to do good, others will use AI to do evil.
The last time our planet faced such an existential crisis was through the development of nuclear science. Sadly, America was the one nation crazy enough to use it.
Click here to access the Scientific American article.
Note: Below is the “Evil AI” image that was created using ChatGPT.
