HSE International ISSUE 109 - Page 45

A rtificial Intelligence, or ‘AI’ as it is often abbreviated, is a term that has received increased public attention in recent times. No longer is it restricted to the minds of imaginative science fiction thinkers – AI is now science fact. But what exactly is it? The arrival of the Information Age resulted in the rapid advancement of humanity through computerization. In the span of mere decades, we’ve gone from supercomputers the size of rooms to ones which can fit in your pocket. Computers were always tools we could manipulate to accomplish complex tasks, similar to using a spanner to change a tyre. Through robotics, we could program machines to change tyres faster and with great efficiency. Artificial Intelligence takes it a step further by allowing machines to learn through failure and experience, just as a human mind would. It wouldn’t be surprising if a machine one day invented a superior method of changing tyres! In essence, Artificial Intelligence is the branch of computer science that seeks to replicate human intelligence via machines. An AI may be able to observe, learn, solve problems, and improve, just as a human would. Through machine learning, an AI can ‘think’ in a similar manner as a human being in order to achieve specified goals. A rtificial Intelligence holds its own sub-genre in science fiction and is usually represented in one of three basic forms – dominant, in servitude, or (rarely) neutral. Of course, it is the malevolent AIs who seek the destruction of mankind which are often most featured in books, movies and video games. While AIs in fiction are almost always created for the purpose of benefitting mankind, things go awry and the opposite is achieved. SHODAN, the AI in the highly acclaimed video game System Shock, was created to manage a space station, but her ethical limitations were easily overridden by a single hacker. In 2001: A Space Odyssey, HAL 9000 malfunctions due to programming conflicts and attempts to exterminate the crew of a space craft in order to achieve mission directives. Perhaps one of the most popular representations of Artificial Intelligence is Skynet in the Terminator franchise, where a war between machines and humanity is initiated by the former as a means of self- preservation. In The Matrix, humanity is trapped in a virtual reality after losing the human-machine war. In the slightly less apocalyptic setting of Ex Machina, Ava, a humanoid machine with Artificial Intelligence, outsmarts humans in the quest for freedom. Fiction writers create stories that may seem beyond reason or plausibility, however many works of fiction have predicted actual modern advancements. Visionary authors such as Jules Verne and H. G. Wells wrote about fantastical technologies that nowadays, are seen as the norm. Wells predicted war tanks, while Verne wrote about space travel a century before mankind ventured to the moon, and submarines in Twenty Thousand Leagues Under the Sea. Credit cards, automatic doors, communication satellites and robots were all envisioned before becoming part of everyday life. Seemingly impossible technologies like teleportation and virtual reality are also on the verge of equaling their fictional representations. Is it then impossible to think that AI can also follow a similar pattern? “Jump ahead a few years and we may have self- operating forklifts and cranes that can safely perform lifts without exposing humans to falling hazards.” B usinessman and visionary inventor Elon Musk is very wary of the looming threat of an AI apocalypse. Musk, best known for his work with Tesla, a solar, electric energy and automotive manufacturing company, expressed his concern for AI running amok. He believes that AI is advancing at a far greater rate than people realise and can easily turn against humanity. Interestingly, Musk is the founder of OpenAI, a company dedicated to creating “friendly AI”. Many other organisations have begun the AI race, including tech giant Google with their acquisition of DeepMind. B y excelling in areas once dominated by human intelligence, AI’s significance is becoming undeniably apparent. Google’s AlphaGo, a division of DeepMind, recently defeated the world’s best player in the ancient Chinese game of Go. OpenAI took only two weeks to surpass the years of experience of professional players of Dota 2, a competitive multiplayer video game. Not only are these perfect examples of the ability of AI to learn, adapt and dominate, but these high profile accomplishments ensure that the future will be shaped by machine learning. Of course, AI’s practical applications are not limited to playing games and shaming e-sports athletes. Google and Tesla both have vehicles with self-dr iving technology, something that was science fiction only a few years ago. While this may be out of reach of the average person today, millions have already been exposed to AI in the form of ‘personal assistants’ like Apple’s Siri and Microsoft’s Cortana. J ump ahead a few years and we may have self- operating forklifts and cranes that can safely perform lifts without exposing humans to falling hazards. Repetitive, redundant work can be completely controlled by learning robots which increase productivity. Hazardous jobs can be completely overhauled and controlled by AI, eliminating exposure to humans. AI- controlled cars, trucks, boats and planes may dramatically reduce travel time and accident rates. Intelligent machines will soon be able to do everything a human can, only with a smaller margin of error and no accompanying muscular or mental fatigue. It sounds like an inevitable utopia. However, there is a looming dark side to this world- changing technology. Recently, Facebook was forced to shut down two AI programs that had developed their own language to communicate with each other. This was the AIs’ solution to solving a challenge, and indicates that thinking machines can easily choose unforeseen paths to problem solving. Tay, Microsoft’s Twitter AI, was also shut down  HSE INTERNATIONAL 45