The Dawn of Intelligent Machines

The World Economic Forum (WEF) reports that by 2022, intelligent machines, powered by complex algorithms, will displace 75 million jobs. But if you’re worried that your future workplace will be dominated by robots or androids, fret not. Offsetting these projected job losses is a more optimistic prediction that intelligent machines will also create 58 million new jobs. The WEF further estimates that humanoid robots (the ones that feature in many science fiction movies) will be adopted by about 23% of companies by 2022, the chief adopters being the financial services industry [1]. Stationery robots have already revolutionized manufacturing lines, with the automotive industry being the primary beneficiaries. Toyota, the world’s largest car producer, produces about 20 cars a minute – its total production of more than 10 million cars in 2017 made possible using stationery robotic arms. Powered by Artificial Intelligence (AI), technology has also made great strides for in aiding medical research. Along with access to Big Data (enormous datasets consisting of numerous data points), AI has helped us better diagnose and recommend treatments for a range of physical ailments. Researchers from Stanford University, for instance, have programmed, or more specifically, ‘taught’ software (through a process called machine learning) to identify cues of depression. The team programmed the software to capture both users’ facial expressions and spoken language. Using these key indicators, the team were able to accurately identify users with major depressive disorder more than 80% of the time [2].

What Defines Us Humans

Promising developments such as the one by the Stanford team give hope for how technology and AI can help better our health and well-being. By 2025, it is estimated that there will be broad-scale use and integration of AI into our healthcare systems [3]. We are also seeing an increasing use of social chatbots for mental health purposes. Social chatbots are AI-powered chat programs that respond very much like human respondents do. One such program is Woebot, programmed to use cognitive-behavioural therapy (CBT) to respond positively and helpfully towards users experiencing depression or anxiety. The specific field of affective computing (also known as artificial emotional intelligence) aims to replicate the human ability to recognize, express and mirror emotions. And to a certain extent, it is doing a great job of it. Yet, not everyone is convinced. Writing about another social chatbot, Replika, journalist Marie Boran calls the program’s ‘emotional intelligence’ as “psychopathic: fake charm that at times borders on creepy” [4]. And perhaps she is not the only one among us that feels that way. Another survey showed that American consumers are hesitant to, or simply feel creeped out from interacting with robots [5]. Part of this discomfort may stem from these intelligent systems seemingly behaving like humans – but just a little bit ‘unhuman.’ Boran refers to this as the ‘uncanny valley of the mind’ – borrowing the term from the phenomena (uncanny valley) that shows how we are fearful, disgusted, or put off by robots that look, act, and maybe even express emotion like us humans – but are just 5, maybe 10% ‘un-human.’

Such reactions are of interest not just to roboticists, but also to psychologists. In one study by Nick Haslam and colleagues, robots are perceived as being comparable (if only a little less capable) in terms of cognitive ability than humans. But where robots were most deficient, however, was in their ability to express emotion – and importantly feel, alongside us. Robots lacked emotional ability, evident in their stifled, static, artificial empathy [6]. What this study tells us is that one part which defines us humans is the ability to connect, authentically, meaningfully – naturally, with our emotions. At time of writing, AI may be able to replicate empathy to a certain extent. It can identify emotions in the other (a critical process and antecedent of empathy). It can also, using its algorithms, make sense of the situation we are relating to them – they can engage in perspective-taking. But perhaps where technology currently lacks is its ability to also feel – to experience the emotions we are experiencing. This essential aspect of empathy (sometimes called affective empathy) is crucial – but not evident in most of our AI-powered technologies. And that leads to a slightly broader question: If AI can identify and recognize both our emotions, take our perspective and then subsequently expresses emotions it thinks is an appropriate response, can we really be certain that it too, feels like we do?

What Artificial Intelligence Won’t Replace Anytime Soon

Empathy – the ability to willfully, and authentically feel alongside another is something AI still finds difficult to replicate. Authentic empathy – the kind that fosters validation, understanding and affirmation of the other, is not something that can be easily programmed. This hasn’t stop roboticists and programmers from trying, however. Japanese company Softbank, for instance, released Pepper, a claimed emotionally intelligent companion robot in 2015 – touted as a possible solution to the country’s aging population and increased rates of loneliness. Despite initial anticipation over Pepper’s possibilities, the robot has been met with mixed responses; some have criticized Pepper as lacking a refined understanding of human emotion and responding in a manner that seems unnatural. Social chatbot XiaoBing, developed by Microsoft, conversely, has seen some success. Using Big Data and machine learning programmed to acquire an understanding of human social interaction, XiaoBing has won over legions of fans – some of the users have even confessed their love for the social chatbot. But the future, according to author Geoff Colvin, belongs to humans who can use the deepest aspects of their humanity to do what machines can’t. The author of Humans are Underrated predicts that machines will never authentically or meaningfully replicate our ability to collaborate, build relationships, use humour, craft stories that rouse emotions. And all these, he claims, stems from our ability to empathize; to feel alongside and resonate emotions with others. How we empathize with ours, and others’ emotions could be the one aspect that not only allows us to hold on to our jobs, but to also hold on to our humanity in an increasingly AI-dominated world.

Note: This article contains excerpts from the forthcoming, “Emotions: A Brief Introduction,” to be published sometime early 2020.