“Artificial Intelligence” is more Like a Myth from Science Fiction Films

Despite the prominent usage of artificial intelligence in widespread technologies, such as Siri and Alexa, to perform daily tasks like adding events to the calendar, sending messages, etc. not many of us can provide the exact definition of the term ‘artificial intelligence. In fact, some people might not realize they are currently using AI-powered systems, as ‘artificial intelligence’ sounds more like a myth from science-fiction movies rather than reality. As once told by McCarthy, one of the founders of the artificial intelligence discipline, ‘As soon as it works, no one calls it AI anymore’ (Meyer, 2011).

One of the possible reasons for the struggle to define the concept is that ‘artificial intelligence’ can be defined both in terms of cognitive science and engineering. Defined from the perspective of cognitive science, artificial intelligence is the ability to process information in a way similar to that of a human being. In this sense, AI features cognitive architecture. One of the main goals of cognitive architecture, according to Techopedia.com, is to define the learning and performance mechanisms of the system, helping to identify the necessary infrastructure for an intelligent system (2018).

According to the same source, the difference between traditional artificial systems and cognitive architecture is the cognitive architecture’s ability to learn based on social interactions, a bio-inspired technique. ‘Artificial intelligence, defined in terms of engineering, is the systems’ ability to execute tasks, which currently require human knowledge. While the task theory of artificial intelligence is not defined yet, Thorisson et al. (2016) determined three things that might be in flux in each system: the task, the environment, and the agent itself. The task not only defines the goal to be achieved but also the situations that should be avoided.

The agent’s perception, goal structures, memory is shaped by the task and the environment. An agent consists of two parts: the controller or AI system, and the ‘body, which sends the signals from the environment to the controller. The term ‘artificial intelligence’ defined from an engineering perspective, can be further divided into ‘strong’ and ‘weak’ artificial intelligence. The distinction between ‘strong’ and ‘weak’ artificial intelligence lies in the fact that ‘weak’ AI can only simulate every aspect of the human mind employing the computer software and hardware, while ‘strong’ AI, employing the same means, can emulate or be equivalent to the human mind ('Artificial Intelligence: Weak AI vs. Strong AI', 2018).

In other words, ‘weak’ AI responds to a human by using a pre-programmed set of operations, while ‘strong’ AI genuinely understands the meaning of the words, and the cognitive processes of ‘strong’ AI strongly resemble those of a human brain. A task as simple as dishwashing can be used as an example to explain the difference between the AI defined in terms of cognitive science and AI from the perspective of engineering. In the first case, AI would involve navigating the identical environment, the kitchen, as humans, recognizing the dishwashing liquid and sponge as the equipment necessary to achieve the task, the way humans do, and achieving the goal, washing the dishes, using the appropriate means. In the second case, the similarity of AI’s cognitive processes to those of a human being does not matter. Particularly, the emphasis is put on the accomplishment of the task itself.

Much debate has arisen in recent years on the topic of artificial intelligence, with one of the most prominent figures in modern science, Stephen Hawking, saying ‘AI is likely to be either the best or worst thing to happen to humanity’ (Hern, 2018). Perhaps the most evident concern with AI could be the danger of unemployment due to the increasing prominence of AI. The fear of unemployment due to the automatization is nothing new: Queen Elizabeth I refused to patent the sewing machine, stating that: ‘It would assuredly bring them to ruin by depriving them of employment, thus making them beggars’ ('Accidental Inventors And The Impact Of Technology On Work And Jobs', 2018). With the gradual advent of AI, not only the tasks in manual spheres (mainly requiring a physical effort) but also those in the non-manual spheres (mainly requiring an intellectual effort) are being performed by AI-powered machines.

This is supported by the fact that there has been significant progress in automating ‘white-collar’ work (Nils J. Nillson, 1984), involving tasks as decision-making, coordinating and communicating, and fact-gathering. A more recent study estimates approximately as much as 47% of US employees are at risk of losing their jobs in the next one or two decades due to computerization (Frey & Osborne, 2013). Specifically, sales occupations, such as cashiers and counter clerks, which do not require a high degree of social intelligence, are in the high-risk category. The low-risk category includes jobs, which require a high degree of social intelligence (ex. business and finance spheres) and those requiring creativity and the development of new ideas (ex. science and engineering). The central question about the above-mentioned fears remains:

‘Are these worries well-founded?’. In this paper, I would like to put emphasis on the reasons I do not believe in the prevailing unemployment due to the increasing prominence of AI. With respect to job losses due to AI, according to the World Economic Forum, it is estimated that 75 million jobs may be displaced by the shift of labor between humans and machines (2018). However, 133 million new jobs, emerged as an adaptation to this new division of labor between humans and machines, should not be disregarded. Therefore, a net gain of 58 million jobs is predicted in the period of the next four years due to the advancement of artificial intelligence.

Moreover, as the numbers of AI-powered systems will escalate, so the demand for jobs in the spheres of AI development, testing, programming, and maintenance, as the machines are not yet capable of running smoothly without any human intervention. In their analysis of 1000 large companies, which adopted AI-powered systems, Wilson et al. (2017) identified three new categories of jobs, in the spheres of business and technology, created to assist AI, explain AI, and guarantee the safety of humans or minimize the risks of AI: trainers, explainers, and sustainers. The first category, trainers, will teach AI how to ‘behave’; more specifically, aid to understand the subtle nature of human communication.

According to the authors, a New-York based start-up, Kemoko has developed a novel machine-learning algorithm to help AI-powered assistants, such as Siri and Alexa, to respond to humans with empathy. The second category, explainers, will help to explain the sophisticated machine-learning algorithms to nontechnical professionals. For instance, algorithm forensics analysts will help to identify the mistakes of AI-powered systems and the causes, resulting in immediate correction. Professionals in the third category, sustainers, will help to ensure that human values are upheld in the performance of the AI systems.

One of the jobs, possibly emerged in this category, the ethics compliance manager will ensure certain biases, such as those in credit approving algorithms, discriminating against people of certain ethnicity or professions, will be identified and eliminated. Moreover, a study by Capgemini Digital Transformation Institute of 1000 companies implementing AI describes the opportunities driven by AI: KLM, the Dutch airline company has been able to improve its customer service efficiency by 35% by employing AI-assisted human agents. Similar benefits, originated from the increasing prominence of AI, could, in turn, lead to the further development and expansion of businesses, creating more jobs (2018). Certainly, as supported by several recent studies, AI can pose many jobs at risk. According to the World Economic Forum, routine-based, ‘white-collar’ works are likely to be automated in the period of the next four years (2018). However, certain jobs, due to their nature, are highly unlikely to be automated, at least in the following five decades. These professions may include those in the spheres of healthcare (ex. therapy, surgery) and education.

While AI-powered systems may excel humans in tasks, such as calculating, organizing big data sets, finding relationships in a sophisticated data analysis, empathy, feelings, or facing novel situations are not their forte. Machines can outshine humans at ‘frequent, high volume tasks’. For example, a professor can read 10.000 essays in one’s 10 years long career, while the AI-powered machine can read millions of them in just one minute; the ‘winning’ algorithms were able to match the grades given by the professor. Likewise, AI systems lack the ability to quickly adapt or respond to novel situations as they need to learn from large volumes of previous data to draw certain conclusions (Goldbloom, 2016).

A task, such as developing a business strategy, which requires the development of innovative ideas, would be accomplished more effectively by a human than an AI. Additionally, AI is not (yet) able to experience and/or understand human emotions. TopTenz (2018), in the list of 10 jobs irreplaceable by AI, specified psychologists, as AI lacks empathy; this is further supported by the fact that mental health and substance abuse social workers have only a 0.3% probability of being automatized (Frey & Osborne, 2013). Not only do they lack human feelings, but AI also cannot convey ideas in a way understandable to people. Zack Thought, a full-stack programmer and a big fan of the Game of the Thrones taught AI to write the next chapter in A Song of Ice and Fire series by George R. R Martin.

TopTenz quoted the novel to demonstrate the AI’s overly complex language: ‘This dragon does not say we had four of a band, or no men or rats and two singers…’ ('10 Jobs Artificial Intelligence Can’t Take Away From Humans', 2018). As humans are the creators of AI, they will be able to control AI-powered machines or collaborate with them, rather than bolstering the competition for jobs. More specifically, three ways, in which AI systems can expand human capabilities have been identified: amplifying, interacting, and embodying. AI systems can amplify or boost human abilities in decision-making, ‘providing the right information at the right time’ ('How Humans and AI Are Working Together in 1,500 Companies, 2018).

According to Jarrahi (2018), AI has outshined humans in difficult tasks with superior quantitative and analytical capabilities. AI will sift through the data to underline the most interesting trends, and human managers will be able to make decisions based on the conclusions from the data, instead of leaving the decisions to be made by AI. AI can innovate human-machine interaction by providing customers with a more effective service: a Swedish bank is already adopting an AI-powered agent named Aida to provide service to millions of customers. However, even though she is able to ask follow-up questions to clarify the issue and recognize the tone of the voice, she escalates to humans in 30% of the cases. To provide better customer service in the future, Aida can monitor the interaction between the human agent and the customer.

Thirdly, AI helps humans in ‘embodying’: ‘cobots’ or context-aware robots powered by AI system, are working alongside humans in labs, factories, and warehouses. Specifically, robots are being widely used in the auto assembly sphere, performing tasks involving repetitive processes, such as assembling and screw-driving, especially for high precision and/or quality, and dangerous tasks. Contrary to the popular belief, the chances of all or nearly all human workers being replaced by AI systems are very low. AI systems are predicted to create more than 50 million jobs in just a period of the next four years. Certain jobs, due to their nature, such as those in the spheres of counseling and the arts, are not likely to be replaced by AI in the next decades. Moreover, instead of entering the race for jobs with AI, humans will be able to control or collaborate with AI systems, enhancing human capacities and providing more opportunities for business growth. 

Cite this page