There was a time when proto-humans must have stood around gaping at the fire, fully believing that the yellow, flickering substance was magical. As time progressed, understanding increased, and fire became a priceless ally. Flash forward a few centuries and humans are pointing at artificial intelligence (AI,) wailing about how it will destroy humanity.
Of course, it doesn’t help that numerous movies and books have depicted the downfall of humanity at the hands of sentient robots. The reality is that AI is far more simplistic, a great deal more limited, and dramatically less dangerous than the uneducated want to believe.
What Is AI?
AI, algorithms, and machine learning are commonly thrown around buzzwords. While all are part of the same technology, they are distinctly different. An algorithm is a set of instructions, written as code, followed by a computer program. For example, a set of instructions could be written to help someone make coffee.
AI would be the same algorithm, but with added, adjustable variables. For example, based on the outcome of a cup of coffee, more or less sugar could be added. The input would be someone suggesting that they would prefer more or less sugar or coffee.
Machine learning is AI, but uses a different, broader method. In this case, the system would be given a cup, kettle, coffee, and sugar, then a cup of coffee would be trial and error created over countless attempts.
AI Learns By Itself
Given that we’ve just said machine learning can “figure out” a cup of coffee, this must mean AI can learn by itself, right? Yes, and no. Machine learning would have to be geared specifically toward making cups of coffee in the first place, by an engineer.
The trial and error process can be seen as learning, and is regularly promoted as such, but is really a set of pre-written instructions running repeatedly. The key is that the system can automatically adjust based on the outcome until success is achieved. It isn’t so much AI learning as it is incredibly smart engineers knowing how to write great algorithms.
AI Will Rule The World
The biggest myth of all is that AI can somehow go haywire, gain sentience, and rule humanity. AI algorithms are very literally unable to do anything outside of what they have been written to do. An AI written to promote an online casino sign up bonus to those most likely to accept it, for example, is completely unable to do anything else. Take a fish out of the ocean, drop it on the moon, and see how quickly it takes over humanity.
AI Is Dangerous
Self-driving Tesla cars have crashed and caused fatalities, and in this regard AI is dangerous. The more faith is put in AI in dangerous situations, the more likely it is to fail catastrophically. To put it another way, AI is as dangerous as humans are willing to let it be. Thankfully AI is largely not put in charge of life and death situations, for very good reason.