Artificial intelligence that does not have an android or metallic “body” to contain it, increasingly responds with politeness to the phone and speaks to us from smartphone and touch screen in the car… Its success and the serenity with which we no longer wonder who is talking to us is precisely that AI has gradually crept in and benevolently in the things we have and do.
So far, real AI (artificial intelligence) have not been seen in reality closed in robotic titanium bodies – nor synto-skin bodies, with skeletons in who knows what. So far these intelligences have made good vacuum cleaners, won go tournaments and recommended things we might like. Could history change? Will artificial intelligence machines lift their heads (or their heads) and rebel against their slow and naive creators?
It is a discussion with many facets, fueled with passion by (human) excellent intelligences. For example, Stephen Hawking, who sees the potential end of humanity in artificial intelligence, Elon Musk, who thinks that they can lead us to World War III, and Vladimir Putin who (sibillino.
All the vices and virtues of artificial intelligence and what humanity must know to dominate it, in the dossier of Focus 204 (see preview)
But not… As Randall Munroe (programmator, comic bookmaker, robotics expert and writer) pointed out in his funny collection of scientific answers to absurd hypothetical questions, \ what seems to escape many critics about the terminators See Atlas (more forward) to believe…
It is a key to reading shared by an authoritative technology expert, Douglas Heaven, who writes on the NewScientist that of machines does not seem to be a triumphant march to the conquest of the Planet. The past and present of robotics and AI are, indeed, studded by “small daily difficulties” that robots struggle to overcome. To understand the future of artificial intelligence automatons, we must follow them in these very difficulties.
Little bad jobs. Boston Dynamics produces some of the most advanced robots in the world: the founder, Marc Raibert, has shown in a recent TED the range of creatures. On stage were BigDog, the four-legged metal mule climbing the stairs (a bit awkward), WildCat, which can run more than 30 kilometers an hour, and SpotMini, quad robot
Awesome (or terrifying, depending on the “team” you have chosen and your favorite movies), but if you have watched all the video you will have seen (around minute 9) also Atlas, robot (almost) anthropomorph
This happens because moving is difficult, but not moving is even more so: when we are standing, still, our brain constantly communicates to the muscles all the adjustments to make, instantly for a moment, just to keep in vertical position. In this, robots have terrible performances: not all, it is true, some are doing it, but judging by the performance of most of today’s automatons, to escape from bad cars we would just have to get up to the raised floor…
Long before BigDog. There are almost anthropomorphic machines that have a glorious past, such as Elektro, talking robots and (alas) great cigarette smoker, presented by the Westinghouse Electric Corporation at the 1939 New York World Fair. Over two meters tall, he could do some small movement, count on his fingers and say (in English) phrases like “I am Elektro” and “my brain is bigger than yours.”
The naive audience of that Expo went into view, also because a showy circular hole in the robot’s abdomen showed without a shadow of a doubt that there was no deception: no operator was hidden inside the machine.
The operators were out, hiding behind a tent, and depending on the particular signals they received they pressed a button or the other to send a command and have Elektro do or tell the most appropriate thing (as well as having him smoke one cigarette behind the other After that edition of the world exhibition, mankind had more to do for several years. Elektro then reappeared in 1950 (as pure curiosity) and, in 1960, had a small part in the b-movie “soft core” Sex Kittens Go to College (which we have never seen in Italian theatres), demonstrating
Thor, the Hammer of Odin. These are not small stories in a sea of successes: any scientist of robotics has his stories to tell about the failures of robots (unless they retract). For example, Alan Winfield (of Bristol Robotics Lab), says he heard about a robot in a scheduled assembly line to open a refrigerator and grab and give cans of soft drinks to passersby during a fair. The fridge door was blocked: it seems that at that point the car has broken the fridge door but, unable to pull out the can, it pulled behind the entire appliance by shaking it in half air like a giant hammer. You don’t know how it ended and Winfield is a bit of a man-defender invention, but, you know, he’s on the side of the machines.
The map of the pop-calisse.
One that is not on the side of the machines is Jesse Newton: fascinated by promises by technology, to keep the house clean buys a well-known robot vacuum cleaner. At home he has a dog that doesn’t look too good in the gut right in the middle of a night when the robot is on duty and passes on what the animal has freed. This is how he starts to track a path on the floors: when Newton told his misadventure on Facebook with a “cleaning path” scheme it turns out that this pop-calitmatic story is more
Shitty robots or, to put it English, Shitty Robots, is a collection of videos about the stupidest and sometimes hilarious performances of robots and humanoid aspirants, online on Reddit (there is also on When it happens to a robot, at least then you don’t feel guilty!
Curiously, however, machines always seem proud of their (in) failures: they do nothing to hide them (there would even be something to learn from them). Winfield – the one of Odin’s fridge hammer – believes that the videos a little joker on the machines are very useful, because the understanding of the dead ends of the evolution of robots then shows us in another perspective, different from that of Ha To make fun of them presents in short the AI and the robotics for what they are and do: very “human” efforts whose results are limited by very human defects and weaknesses, in other words \ the best proof that ro
But there are also embarrassing and very negative cases. Take the story of Tay, Microsoft’s chatbot, which debuted in 2016 on Twitter. Designed to learn words and expressions in natural language by speaking directly with users on social media, a few hours after the activation he wrote sexist, racist, Nazi comments… Microsoft turned it off within 24 hours. It is a story that reveals the other side of the coin (we).
Facebook incident. From whom did Microsoft’s artificial intelligence learn thoughts and words? ♪ What we really need to fear is not robots, but ourselves ♪ ♪ It’s the hard reflection of Mark Riedl, an expert in artificial intelligence at the Georgia Institute of Technology in Atlanta: ♪ Under
Unfortunately, these are the scenarios that “took” the most on social media and that make audiences like the worst buffalo (fake news, to say the English), as shown by Facebook’s recent experience. In August 2017 FB ended a verbal interaction test between two chatbots because at some point the “natural language” used by software had become incomprehensible: the two had started using very modified words and rules (probably more effective,
It was not a first time, but for some reason on this occasion, and for most media, the event had become worrying (“does artificial intelligence become too intelligent?”): we told this story – in the right tones – in the article The Secret Language of the,