By Raz Koroh
When Intel created a new interface for Professor Stephen Hawking’s monitor in 2014, Hawking’s speech rate was doubled and his productivity was improved ten-fold, according to the technology giant. Hawking himself admitted that science and technology improves the lives of people with disabilities, and that without it he would not have been able to speak to people.
Prof Hawking then prophetically said, “The primitive forms of artificial intelligence we already have, have proved very useful… but I think the development of full artificial intelligence could spell the end of the human race.” Is Hawking correct?
Here are several factors to consider.
Learning curve. When Lee Sedol beat Google’s AlphaGo program on March 13, 2016 to finally make it 3 – 1 in a five match series of the ancient board game Go, denying the artificial intelligence a clean sweep was the most priceless win to the holder of 18 international titles and strongest human Go player in the world. Then in the final Game 5 on March 19, 2016 Lee employed the same strategy that he had used to win Game 4, and lost 4 – 1. AlphaGo had learnt its mistakes in Game 4 and improved it enough to surpass the human champion in the next game. Humans have been honing our collective knowledge of the game for more than 2,500 years, and artificial intelligence did the same in a matter of days!
Open-sourcing. Within hours of Lee’s victory in Game 4, Microsoft publicly admitted that it has decided to use its Minecraft-based AIX platform to open-source and test artificial intelligence projects, saying that they will involve general intelligence similar to the nuanced and complex way humans learn and make decisions. Researchers can use it to, for instance, teach a real robot to climb a hill without having to repair or replace it every time it falls into a river. Computer scientists at a Microsoft Research Lab in New York already tested a Minecraft character to learn climbing a hill; Microsoft had bought Minecraft from Mojang in 2014 for USD2.5 billion.
Hacking. Nowadays, it is not all that difficult to hack into someone’s computer. For instance, hackers can gather as much information about the targeted individual or company as they wish through search engines like Google or Yahoo, or they can simply use hacking software to begin scraping the internet for every piece of information they could, including every social networking account, every email account and every online service the target had ever signed up to. They build a profile of the target. From Facebook bits of information, they can use popular family tree history websites to verify birth details, they can use Twitter to find work address, photos and places frequently visited. Now the profile is more personal. When they send emails to the target, they will include tiny images to fingerprint computers used by the target, to find out operating systems, browsers used, browsers add-ons, and security software used on the computers. Now they are ready to launch their attack by constructing the malware to infect the targeted computers. If they can bypass the computer security, a remote access Trojan can be planted that can begin tapping into everything. A complete administrative control means they can remotely open up applications, browsers, download cookies, control the camera and microphone and everything else. They can install a keylogger and record every password typed into every website visited, go through emails and lift copies of sensitive documents sent, and watch as the target connect to work computer to look at files stored there.
These three factors alone would send a shiver down anyone’s back spine. A weapon of mass destruction system replicated within days by using a freely available off-the-shelf artificial intelligence platform. Imagine what would happen if artificial intelligence falls into the wrong hands. If we are not careful, Hawking’s prophecy may just become a reality.