Stubborn minds are our greatest threat with AI.

 

Our Biggest Fear About AI should not be AI itself

Today's (Sept. 23) Washington Post article in its ‘Economy and Business’ section was titled with a provocative question about artificial intelligence: “Is this how the world ends? Extinction scenarios are taking over the AI debate.”

The near full page article covered lots of ground, including some scenarios far more speculative than realistic. Yet it missed two important considerations—and one seemingly obvious means to reduce the risk of an AI-driven catastrophe.

First our greatest danger may not be AI independently deciding to eliminate humanity. That’s possible, but a far more immediate and foreseeable danger is what humans will do with increasingly powerful AI.  There is no doubt it will assist malicious individuals, organizations, or governments.  They are already doing so.  Its greatest existing capacity is using AI in developing increasingly sophisticated means of mass murder, including biological, chemical, cyber, or other clever means of mass destruction.  This is far more predictable than the notion that AI will suddenly develop a desire or independently exterminate its creators.

Another, very different danger is AI systems designed to help us - making consequential mistakes. A well-intended AI system involved in medicine, biotechnology, infrastructure, financial systems, or public health could produce an error that is amplified at the scale, and speed that AI already operates. Too many times, in human history good intentions led to catastrophic consequences.

Yet another fundamental characteristic of AI deserves even more attention.  It’s dependence on electricity.  No matter how intelligent any AI system becomes, it cannot continue operating without the physical infrastructure that supplies its energy and physically supports its computing systems like water for cooling.

That raises a basic question: Humanity can now start an internationally agreed-upon mechanisms to disconnect, isolate, or shut down critical AI systems in an emergency.

Governments will not create a giant kill switch that turns off the world’s electricity. That would itself cause enormous disruption and potentially cost many lives. Instead, we should consider whether critical AI infrastructure can be designed with independently governed, internationally coordinated emergency shutdown and isolation capabilities.

The ultimate safeguard against an AI catastrophe is not trying to make AI perfectly safe. It should be making sure that humanity always retains the ability to stop or contain it.

Unfortunately, that brings us the biggest obstacle of all. International trust and cooperation.  Technology has never been the real problem.  Its how we citizens and our policy makers decide to use it.

Our species has developed super-extraordinary technological capabilities while remaining remarkably poor at coordinating and cooperating to manage the risks those capabilities create. Hyperpolarization, nationalism, competition, distrust, and short-term political interests preempt us from acting collectively. Even when the danger is shared by everyone.

Yet we are a profoundly social species. And our survival has always depended on our ability to cooperate, coordinate, communicate, and care for one another. Now we need to move this to the global level where concepts our minds created over 10,000 years ago like political ideology, nationalism, economics, religion, and other divisive non-genetic identities.  These now, continue to override our biological imperative after millennia of bloodshed and needless suffering.

The most important AI question isn’t:  “Will AI destroy humanity?”

 It’s: “Can we transform our mind’s thinking and learn to cooperate well enough to prevent humans from using AI—or allowing AI systems—to destroy us?”

AI is becoming the most powerful tool our species has ever created. Our greatest test isn’t making machines more intelligent.  It will be, can we become wise enough to use them by uniting as the one race/species we are.  The human race.

Technology determines what’s possible.  Our minds cooperation will determine what happens next.