NVIDIA defines machine learning like this: “Machine Learning at its most basic is the practice of using algorithms to parse data, learn from it, and then make a determination or prediction about something in the world.”
Instead of preprogramming software to complete a specific task, as narrow AI does, machine learning uses algorithms that allow a computer to learn from the vast amounts of data it receives so it can complete a task on its own.
Intel puts it this way: “Machine learning is this idea of marrying algorithms and statistics” so that the machine can learn from new data.
NVIDIA uses machine learning to allow sensors, cameras, and computers to process the images they see and create a type of computer vision.
The hardware and software work together to understand what the edge of a road sign looks like, what letters of the alphabet are, and how to identify where an object begins and where it ends. When all of these algorithms work together and learn based on what they see, they identify things like stop signs for driverless car technology.
There is nothing stopping these idiots from reusing this creepy serial-killer as an “adversarial machine learning” opponent/trainer for another AI like, for instance, training a “detective” AI. (see here: https://mascherari.press/introduction-to-adversarial-machine-learning/).
The problem is that while the detective AI is getting better after being spoofed billions/trillions of times by the serial killer AI, the serial killer AIS getting better as well after each of its failures. There are many scenarios that the serial killer AI could be used legitimately or nefarious. Repeat this process over and over for say 20 to 30 years along with exponential growth computer power and everything being increasingly connected to the Internet and feeding data from the real world to cyberspace (like Alexa the voice assistant spy, smart utility meters, government spyware in your cars, planes, smartphones) and there will come along a problem for everyone.
I wonder what a neophyte serial killer be willing to pay to have a captive serial killer AI plot all his moves?
Infinite thanks! That was just what I needed to understand this.
All of my work has been in analyzing poetry by computer with the words viewed as symbols, not as AI contextual content. I never got into AI. My field was computer programming languages so, again, symbols. Obviously, I need to expand my reading. Deep thanks.