I remember how excited I was back in the days whenever a science-fiction movie was played on the dusty TV box set at home. Battles between intergalactic universes with illuminating swords, fancy wings of a flying car traveling through time: those images are the indispensable part of my childhood. You might have recalled the video-call technology similar to that of Skype in Star Wars, or some deviants of the current hoverboard that was previously created in the Back to the Future franchise. Conversely, there are some instances of technology being described as dangerous and damaging, and sometimes are portrayed as the culprit to the human existence. On this account, Skynet, the fictional neural net-based program of the Terminator, might be the most famous victim of them all. The lively scene of the doomsday in the movie has, more or less, inflicted upon us a questionable and doubtful stance on the technology that it is based on. Artificial Intelligence (AI) is treated with extreme caution, as the reaction to the development of AI is vastly contrast to that of what we usually take on other innovations.
As a response to the widespread public concern about AI, some experts have already proposed on developing a regulatory framework for it. Elon Musk, the founder of Tesla, has been extremely vocal in expressing his view on the preemptive measures in the dealing with artificial minds, while the Facebook CEO Mark Zuckerberg shows that he is less worried or concerned about the exponential growth of the industry. Still yet, no side has taken the absolute victory. Could it be too early to address this problem now?
So, what might you have missed?
One of the first astounding achievements of AI might have been founded on the chessboard. The first of its kind dated way back in 1998, when Deep Blue, an IBM development won a standard chess game with a World Chess Champion. In 2015, a machine called Komodo won a series of handicap matches against 7 of the most prominent chess players of the world. In addition, with the highly complicated board game of Go, the Google’s project Alpha GO challenged a human professional player to a five-game series, in which it won four in 2016. It swiftly sealed the victory against the best GO player a year later. The chains of achievements did not stop there, as it quickly stepped into (competitive) gaming. After a year of continuous “practice”, from a complete neophyte, OpenAI easily dominated one of the most proficient players in the game of DotA (only a one-on-one contest) in front of millions of viewers spectating worldwide.
What is AI actually capable of doing, now?
Artificial Intelligence, is the virtue of machines in expressing its “intelligence” by the process of learning and problem-solving of in a repetitive manner. Their superiority to human in this aspect is that as a machine, they work tirelessly. As much as our intelligence is used to find the most favorable way to deal with our own paradoxes in our everyday life, the objective of an AI machine is to optimize their own outcome in that matter which is specifically defined in the environment to which they are exposed.
From these achievements, we could get easily deceived that artificial intelligence had been making tremendous progress as we previously experienced with the Internet. As a matter of fact, the scope of its development has proven to be much less impactful to our everyday lives than many may claim to be.
To understand how much of an impact that is, let’s take a look at some of the famous achievements of AI. When you send your postal mails, with the help of AI, a machine can now sort in which “box” a certain mail should go to by looking at the handwritten postcodes on the envelopes. A similar function can be seen on your electronic mailbox as well, as it will try to sort out spam or emails with promotional content into other sections of your mailbox, rather than going directly straight to your inbox as it was 10 years ago. Some websites, such as Amazon will show relevant recommendations that you might be interested in after you have finished the purchase for one item. And oh, who could have missed out on the face recognition feature of the newly-released iPhone X just last week, which is also constructed based on a function of AI!
If you are looking a single entity that focuses on developing artificial intelligence, look no further than Facebook. In retrospect, Facebook is actually a large condominium of various AI technologies. From friend recommendations, face recognition in pictures where you could tag your friends, to just simply how adverts on Facebook can cater to your own preferences, these minor features could support you to make better decisions, but they cannot directly replace you in making your choice. It could influence your decision though, as some might have speculated that the result of the 2016 US Presidential Election was swayed by the involvement of artificial intelligence – however, it has to have the external influence from the humans themselves who create the algorithms.
The quantitative impact of AI
Much of the growth in AI that is discussed nowadays probably would only end up in speculation, since it is extremely difficult to directly quantify because of such uncertainty in the growth of the industry as a whole. We could actually divert this by calculating the impact of AI on some other intermediate factors which can be translated somewhat to the overall impact on the economy. For example, Accenture estimated that the growth of AI could potentially double the global economic growth by 2035 (to 4.6 percent from a 2.6 percent baseline projection) if the existence of AI technology could increase labor productivity by 40 percent. The boost in labor productivity, as is mentioned in the article, can be helped by the fact that AI helps workers to find more efficiency in their use of time and instead focus on developing and innovating other products rather than devoting their effort to complete less meaningful tasks.
The other economic indicator that attracts attention from the public is on how the impact of AI could impact their employment. Usually, the development of one certain technology could an adverse effect on a certain industry, but at the same time the existence of “new” jobs will take over. Take the classic example: with the existence of the car, it reduces the demand for horses, but it opens many more job opportunities in other areas such as car manufacturing, car maintenance and so forth. So when the impact is neutral in terms of employment, the overall matter could only be simply a reallocation of jobs from one sector to another. It could be inferred that the net positive effect can be more visible as a technology evolves in the long run. For example, with the upcoming existence of automated cars, it is likely that taxi drivers could lose their job, but on the other hand, high-end technical experts for developing and maintaining the system will get it. Fikey and Osborne (2013) believe that about approximately just under half of the working population would be affected by the increasing adoption of AI on an occupational basis, but luckily, there is only 9 percent of job tasks that can be automated (Arntz, Gregory, & Zierahn, 2016). That is true since the machine could only tackle a problem based on the number of instances that they are exposed to and requires a massive database, so the job occupation that is characterized by novelty and creativity could not be replicated by machines. For example, for marketing jobs, the machines will not be able to outperform humans since doing a marketing campaign requires originality, in order for it to stand out of the crowd. There will be some jobs that are more vulnerable to it, but some will not. So, actually, the fear that job replacement is not actually that daunting as many have described, or imagined.
Of course, we have not yet understood the comprehensive picture of the development of AI since there are many ways that they could place an impact on our economy, and we are only in the initial stages of understanding the technology. As it starts to unveil themselves in the near future and discussions are facilitated between involved parties, we would definitely get better predictions on the economic impact.