Artificial Intelligence: The integration of machine learning and dispelling of myth
Artificial Intelligence and Machine Learning are not relatively new concepts. Having popped up in countless science fiction novels and films throughout the decades the idea is quite old, but in recent years humanity has made leaps and bounds in creating self-learning and self-managing programs and machines. There has been a lot of skepticism and fear surrounding artificial intelligence, some of this coming stemming from films, novels, and video games such as the “Terminator” franchise, “2001: A Space Odyssey”, and the “Mass Effect” video game series. So far, these fictional representations are just that. Fictional, and relying off nothing more than hypothetical situations (Brooks, 2018).
The process of creating artificial intelligence has taken many decades, with the process to create an artificial intelligence being arduous. Originally, artificial intelligence were created using a set of rules, with the creators having to write lists of rules for the AI to follow. This was a very inefficient way of creating artificial intelligence and a solution was needed. To improve the learning capability of artificial intelligences, an algorithm was created to replace the lists (Zadah, 2018). This algorithm is to let the program learn and assist in whatever function, program, or machine it must run. Machine learning can take various forms; whether from running various scenarios to find a solution, learning from its mistakes, or improving efficiency based off what knowledge it can acquire tests it can run.
Machine learning is currently a rapidly growing technological field (Jordan and Mitchell, 2018), and is split into three main research areas (Michalski, 1984). These three areas are task-orientated studies, cognitive simulation, and theoretical analysis (Michalski, 1984). Each of the three areas are connected, with research in one area tending to lead to another, subsequently with progress being made in both. Through the breakthroughs and achievements in these areas, machine learning (and by proxy artificial intelligence) is improved. As stated earlier, machine learning is only one part of artificial intelligence.
In an increasingly digital age, artificial intelligence has gained more prominence and importance. More things are handed over to these programs to be run, however these programs are still limited, even with machine learning. An artificial intelligence designed to make other machines cannot currently play a game of chess (Hutson, 2017). Something rather important to note about AI is that it is still in development, and it will most likely be decades until an artificial intelligence that can think like a human (or better than) and can run various different systems and programs will be created (Hutson, 2017) (Brooks, 2018). This however has not stopped various companies and agencies from using AI, believing that it can improve different applications (Copeland and Reynoldson, 2017). An example of this is the Australian Defence Force studying the military use of artificial intelligence; this would be targeted towards relieving personal of roles where fatigue and speedy decision making are important (Copeland and Reynoldson, 2017). However this does not mean that a real life interpretation of “Skynet” is being built anytime soon.
Despite any fears surrounding artificial intelligence, researching and development into it has progressed and proceeded. With the addition of machine learning, AIs have now become more powerful and increasingly helpful in running and managing programs and software for whatever their creator or buyer needs. In conclusion, though it may take many more years of hard work, artificial intelligence may go from being the work of science fiction, to the work of real science.
Resource:
Brooks, R. (2018). Robotics pioneer Rodney Brooks debunks AI hype seven ways. [online] MIT Technology Review. Available at: https://www.technologyreview.com/s/609048/the-seven-deadly-sins-of-ai-predictions/ [Accessed 24 Apr. 2018].
Copeland, D. and Reynoldson, L. (2017). Pandora’s Box – How to avoid ‘summoning the demon’: The legal review of weapons with artificial intelligence (Humanities & Social Sciences Collection) – Informit. [online] Search.informit.com.au. Available at: https://search.informit.com.au/fullText;dn=371832332464102;res=IELHSS [Accessed 18 Apr. 2018].
Hutson, M. (2017). AI Glossary: Artificial intelligence, in so many words. [online] Science. Available at: https://pdfs.semanticscholar.org/4924/45d775f5fec43face6aa8a6c78fb48407f61.pdf [Accessed 16 Apr. 2018].
Jordan, M. and Mitchell, T. (2018). Machine learning: Trends, perspectives, and prospects. [online] Science. Available at: http://science.sciencemag.org/content/349/6245/255 [Accessed 18 Apr. 2018].
Michalski, R. (1984). Machine learning. Berlin [u.a.]: Springer, pp.3-4.
Zadah, R. (2018). The Difference Between AI and Machine Learning. [online] Intel. Available at: https://www.intel.com/content/www/us/en/analytics/ai-luminary-reza-zadeh-video.html [Accessed 24 Apr. 2018].