Artificial Intelligence has been an endless source of fascination for decades, prompting moral and philosophical debate on what it means to sentient or self-aware, and what rights should come with those states of mind. We see potential quandaries fictionalized in popular TV and literature, from Data the loveable android in Star Trek: The Next Generation to Samantha the Operating System in the hit movie ‘Her’. What is frightening to consider is exactly how close we, in fact, are to sharing a world with these seemingly futuristic A.I.s. Could a self-learning and emotion-feeling robot make an appearance in our lifetime?
As evidenced by the recent artificial intelligence advances listed below, it is a possibility we would do well to consider.
Imagine a technology that could tell what you’re feeling just from looking at your face. The newly developed ‘Emotient’ software records your facial expressions at 30 frames per second, then translates its findings into an interpretation of your emotions. Perhaps this doesn’t sound too scary, considering we could simply conceal our emotions through a neutral facial expression, thereby fooling ‘Emotient’. This is harder than you might think, argues Emotient creator Ken Denman. The technology picks up on the smallest of ‘micro-expressions’ that even humans tend to miss, therefore seeing straight through a poker face and proving accurate in over 90% of its readings. This highly advanced development in what could technically be used as lie detection is somewhat disturbing. Will interrogations of the future be carried out by intimidating robots?
Next on the list of artificial intelligence advances is something equally disturbing. In Carnegie Mellon University in Pittsburgh, a group of roboticists have helped a robot learn how to grip, not through programming, but through letting the robot teach itself. The team gave the robot arms which it could extend, rotate and use to grasp objects. The robot knew that the objective was to pick up an object (an act it would detect through the sensors in its clasps). The robot was then left to its own devices for hours, clasping up to 50,000 times as it tested different ways to achieve the objective. Every time it failed or succeeded in picking something up, the robot would become better at predicting which approaches were more likely to work. Without any human intervention, the robot has now developed past the stage of clumsily gripping nothing but air for hours on end, to successfully grasping things over 70% of the time. It may seem like a small step, but how long until a robot not only knows how to learn but also starts deciding what they want to learn?
The experiment recently conducted at the New York Polytechnic Institute has demonstrated an incredible milestone for robotics, and moving us into dangerous territory as a result. In what might be one of the creepiest artificial intelligence advances ever, a robot has passed a self-awareness test it was initially thought only humans could pass. Three robots were used in the experiment, all of which were programmed to know that two of them had been given a “dumbing pill” which would make them mute. When the professor then asked, “which one of you hasn’t received the dumbing pill?”, only one robot would have the ability to answer “I don’t know”. This is exactly what the speech-capable robot did – he responded that he did not know which of them had not been given the dumbing pill. Upon hearing its own reply, however, the robot deduced that it must be the one who avoided the dumbing pill. The robot, in other words, identified itself as a separate entity from the other robots and was able to make a logical conclusion based on this.
A mind-controlled prosthetic limb has given a young, paralyzed man the opportunity to regain some normality back into his life. The fingers of his robot hands have been fitted with sensors that, when touched, induce electric signals that mimic touch sensations. Blindfolded, the man can determine which finger is being touched close to 100% of the time. Incredibly, the man can also control the prosthetic arm through thought, thanks to the electrodes that have been inserted into his sensory and motor cortexes. The technology is clearly being used for a fantastic purpose and should give us high hopes as to what more we can achieve within the field of prosthetic body parts.
The implications this has for future robots is far from clear, however. Will robots ever feel touch, rather than just sense it? Could robots ever feel actual pleasure or pain from being touched, using a variation of the aforementioned technology? Even before we start worrying about robots, we should perhaps consider the idea of fitting a human with a robotic arm that is stronger than a biological arm. Should we ever allow a human to be fitted with a robotic arm that can extend further than the human arm can, or which could double up as a useful tool – transforming into a pen, or a knife, when needed? Should we fear the potential danger of this human/robotic hybrid, or welcome it as a logical next step in mankind’s advancement?
As exciting and potentially beneficial artificial intelligence advances could be to the human race, even some experts in the field agree there is plenty of reason to be concerned. Stephen Hawking believes A.I. will end humanity if we are not careful, whilst Elon Musk compares A.I. to ‘summoning the devil’. “Once the machines are sentient and super intelligent… they’ll be able to make even smarter machines. And even smarter. And smarter. And suddenly, humans won’t be so necessary anymore,” writes Drake Baer in Business Insider. It seems the dystopia of science fiction might just be around the corner after all.
If you want to learn more about the recent artificial intelligence advances, consider reading the related article “Five Creepiest Advances in Artificial Intelligence.”