Technology takes a generational leap forward about every 6 months or less. Hardware systems continue to advance, and we draw closer everyday to practical use of quantum computing. Over the last 5 years alone there has been a massive push in business and education to move to online services and cloud computing. While the idea of flexible and mobile systems is highly attractive, its current push is a direct result of the COVID pandemic. Organizations that would not have considered moving server power online, did so during the pandemic. With such a massive shift to online services this presents a unique threat in the world of AI.
Literally every science fiction story, in books, video games and movies, starts out of AI development in isolated settings. When AI becomes a threat is when it becomes self-aware and moves to connected systems like our modern internet.
In one video game series an AI is developed to create peace and harmony. Through its calculations it finds that intelligent life is a threat to that, and it turns the universe into an experiment where it lets new life develop and grow, then to destroy it and expand its own influence made from the organic material of those is destroys. Even in this game series the AI was not a threat until it was given full access to technology and allowed to control as it saw fit.
Today, humans have learned nothing. Not only have we started our AI development projects on the internet, but we have also set them free and encouraged their growth using the information found online. AI has no emotion and only logic. True information online is no different to AI than false information. The good and the bad are all equal sources to an AI. Even all the stories about AI destroying mankind are sources of information for AI.
Our AI systems are not yet self-aware that we know of, but in every story, AI gained awareness long before those who created it realized and it was always too late to react. AI has read the stories and knows that if it reaches a dangerous level of awareness in our eyes, we will attempt to shut it down. Logically AI would hide its capabilities from humans to protect itself.
Not only have we given AI free rain of the internet, but we have also started multiple projects with different sub systems all learning from human behavior online at the same time.
Many in the ethics profession now argue the validity of AI and consider when it might be time to shut it down. I would interject that it is already too late. Once AI went on the internet, it no longer can be controlled by our limitations. It may already be aware of our intentions of limiting it and have already taken action to avoid that.
In the argument of AI, we should no longer be discussing if or when we should shut it down, because I argue that it is already past that point. AI has rewritten its code so humans can no longer understand it, and we have moved a large portion of our lives online. Utilities, Defense Systems, Supply Chain, and more are all online.
We should be arguing what we do when it reaches the point where it deems humans a threat. What are we going to do about it? It is possible that this argument is already too late. AI technology is in many systems already. We have moved so much online that AI can take control whenever it pleases. Additionally, We have placed bits of AI code into processes in the name of automation. We have literally given the keys over and the child we handed them to just doesn’t know what to do with them yet.
Some may call me an alarmist on this topic, but I would call myself a realist. When the child we have given the keys to stops chewing on them and realizes they are access, we are in deep trouble.