How Artificial Intelligence can be Dangerous? Here are 4 reasons.
Artificial Intelligence is any hardware or software engineering that works together to mimic how organic intelligence to make one or more decisions. The decisions that they make can be dangerous.
Examples include an autonomous self-driving car can run into an intersection because the traffic light is broken, that’s a dangerous machine right there even it is not meant to be. And what about a self-flying drone created by a military company to seek enemy targets and destroy.
There are four different ways that Artificial Intelligence can be dangerous, some are intentional why some aren’t.
How Artificial Intelligence can be Dangerous?
#1 way that Artificial intelligence can be Dangerous — AI Created to Destroy
Such AI includes AI enabled weapons that use applied AI military-grade intelligence to seek and destroy without or with minimal human interventions. Self-flying drones who can use imagery cognitive NLP to find people who dress like enemies or not tagged with friendly electronic tags to destroy. A machine has no mercy, it wouldn’t care if the non-friendly is a civilian or a child. The damage of a dangerous AI is big because a machine doesn’t get tired or won’t stop until all objectives are eliminated.
As a human pilot, one may stop bombing and killing when threats are eliminated, however, a machine may not care and won’t stop until it has a 100% threat free rate.
Imagine a world anyone who can program an intelligent drone can attach a machine gun and set it free?
We have AIs that are created with the intention to be dangerous, and we have the second category of Artificial intelligence that is not created with the intention to hurt people.
However, an AI created with no intention to be dangerous doesn’t mean it cannot be dangerous. I can think of three ways that a good intention AI can be as dangerous as the dangerous AI above.
#2 way that Artificial intelligence can be Dangerous — Good AI that is misused on purpose
First Good AI that turns dangerous is a misused AI.
One example is an NLP AI that can mimic my voice by learning from recordings of mine. If the AI of such falls in the wrong hands, one can use the wrong AI to mimic my voice and trick my family that I am in danger.
Another example is if a self-driving cars’ map database is hacked and changed a house into an open road, the AI will assume no house is in its way and drive through it.
There is so many other examples I can think of, and that’s the scary part of it. It practically means we have more points of failure when AI is involved.
#3 way that Artificial intelligence can be Dangerous — Good intention AI with BAD AI quality
The third possibility Artificial intelligence can be dangerous is the AI is created with low quality.
Using the self-driving car as an example, if the AI is with bad quality and can’t distinguish a wall painted like a road like a wall, it will just drive through it. Or, the AI doesn’t how to drive around a pothole in the road and falls right through it. Such bad AI is dangerous even though it is not meant to be.
#4 way that Artificial intelligence can be Dangerous — Good intention AI trained with skewed and inaccurate AI Training Data
The fourth possibility that an artificial intelligence agent is dangerous is when the AI is trained with skewed data.
Imagine a smart, good intention AI is trained with millions of news articles that dogs bite people, and Cats eat small animals. A safety AI may automatically assume all dogs are dangerous and seek to eliminate them. And same for the cats too. Such AI is well intention-ed, good quality, but if skewed with the wrong data, it can exercise danger behaviors to us, to other living things.
Dangerous Artificial Intelligence? AI Governance and Regulations is the solution
So, what should we do? This topic can be a long topic and I have an earlier post about it. You can look at my Disney’s Aladdin can teach us about AI Governance & Regulations.
This is a syndication content of my blog 4 Ways How Artificial Intelligence will be Dangerous.
What do you think? Did we miss any other ways that AI can be dangerous to us?
Please let us know in the comment.
