So you see all these science fiction films chock full of robots that are just as lifelike as humans. Towering machines with such complex machinery that you can’t even fathom how they function. Impossibly advanced rogue Artificial Intelligence nursing plans to take over the world.
You see all these sci-fi tropes trotted out time and again and think, that’s never going to happen in real life! But the truth is, actual AI has evolved much further than most people commonly realize. We may not be at the point of catching up with sci-fi robots just yet, but these following four AI programs appear ripped from the pages of science fiction itself:
Deep Learning Creates Unfathomable AI
Imagine an Artificial Intelligence decision-making program that is so powerful, and so impossibly complex, that even its own creators will not be able to tell you how decisions are being made by the program.
This is the problem posed by deep learning, which is a method of programming computers to receive external data, process the data, and learn to perform new tasks independently.
While this would create Artificial Intelligence that can operate without human supervision, the decision-making process followed by such AI is so complex, and makes use of such a vast network of tiny processors, that even the creators of the AI will be unable to predict what the machine will learn next, or how it will respond to changing external parameters.
For instance, you can design a program that learns to drive cars using deep learning and by studying human drivers. But if one day that program causes the car to ram into a tree or sit out a green signal, you won’t be able to study the program’s internal decision-making process that led up to the wrong decision. And if you can’t figure out how the program performed the wrong function, how do you go about fixing it?
The Robot That Makes Robo-Civilian Rights A Serious Issue
If you ever hear someone talk about how machines deserve the same rights as a human being, you will assume that person is either insane or… actually, that’s the only assumption any logical person would make. Our fridges need rights now? Our computers should be protected from having to be kept turned on day and night?
But the question of rights for Artificial Intelligence is a serious issue in certain important circles, and the debate is opening up even further mostly thanks to the robot known as Sophia. Created by Hanson Robotics, the company claims that Sophia is capable of talking, learning, and feeling emotion like a regular human being.
While many experts accuse Hanson of exaggerating Sophia’s abilities, and it is known that many of Sophia’s unnervingly life-like lines spoken to audiences are scripted, the fact remains that the program has received worldwide attention as the first possible robo-human. For instance, Sophia is the first robot in history to be granted citizenship status in a country, as well as being the first non-human to be given a United Nations title.
Real Life Cyborgs Are An Actual Thing
A cyborg is a popular staple of science fiction and is generally described as a part human-part machine entity. The real world took one of its first significant step towards future cyborgs last year with Elon Musk’s new company called Nueralink. The aim of the company is to create electronic devices that can be implanted directly into the brain. Cruder versions of such implants already exist and are used to carrying out basic medical procedures directly within the brain that help treat various diseases, from Alzheimer’s to Epilepsy.
While Musk states that medical aid is the immediate goal of Nueralink as well, their ultimate ambition is to create electronics that can mesh completely with brain tissue without causing damage. The machines will then allow the user to tap directly into the resources of the internet using only thought commands. Eventually down the line, Musk hopes to see a new race of humans who are directly linked to and able to control the machinery and computer programs around them using the chips in their brains.
The AI That Can Out-Learn Humans
The Libratus computer program pitted its poker skills against the top four poker players in the world last year and ended up beating them soundly. The tournament was played out over a course of 20 days, instead of the usual 13. By day, Libratus competed against the players. By night, it perfected its game strategy by analyzing every step of the gameplay and results. It continued to perfect its strategy over each day, resulting in a steadily improving performance.
At the end of the tournament, Libratus was ahead $1,766,250 in chips, leading one player to remark that the program was so good, it was like playing against someone who could see your cards.
While winning a poker game might seem like a cute stunt, the ability revealed by Libratus to fill in the gaps in its knowledge by studying an increasingly expanding data set and making highly accurate guesses which inform future actions means the program can find use in highly technical areas of operation, including medicine, cybersecurity, and even business negotiations.