The robotics industry is one of the fastest growing industries of the 21st century. Found almost everywhere, robots have integrated into human society. By this, I am not referring to an apocalyptic world in which robots are superior to people, but rather the integration of robots as an aid function in society. Robots can be found online, as chat-functions, algorithm operators, but also physically, in the medical and scientific industries. It is safe to say that human advancement in almost any discipline has been improved due to the helping hands of robots. Arguably the most interesting and somewhat eerie development in recent years is the increased lifelikeness of robots. They seem to have every ingredient to make a human, but they are not made from flesh and bones, and have a computer for a brain. Many of us find these ultra-lifelike robots unsettling, as though one cannot quite put their finger on what seems off – similar to deep-fakes. This uncomfortable feeling has a name, known as The Uncanny Valley, and engineers are unbelievably close to overcoming it.
The Uncanny Valley is best described with a graph, as theorised by psychologists and engineers. Originally, it was assumed that the more lifelike a robot is, the more likeable or attractive a robot is to humans. For example, the well-known robot R2-D2 from the Star Wars franchise, whilst some may say he is cute, it has never been suggested that he is attractive. However, in Japan the first android news broadcaster “Aki Kananasy” was labelled attractive, for her lifelike features. Of course, these are just two examples, but try thinking of some robots you know – the correlation between lifelikeness and attractiveness is undeniable. However, there comes a point where the attractiveness of a highly lifelike robot suddenly drops off. When presented with a robot in this region of lifelikeness, viewers are repulsed by it, as though from a horror film. This is the Uncanny Valley.
In terms of scientific advancement, the Uncanny Valley is an obstacle. Engineers have dedicated decades to the development of robots under the premise that lifelike robots are more attractive. The Uncanny Valley has proven that this is not necessarily true. The theory supposes that this void can be surpassed – that if engineers can successfully create a robot that is so lifelike that the general public cannot decipher between a robot and a real human, then the Uncanny Valley will be over. One may wonder what the benefits of having such a humanoid robot are, but the uses are infinite, most prominently for research. Analysis of human behaviour can be modelled by robots. This makes the process for research much smoother and easier since these robots could be made readily available. Humanoid robots can also be used as test subjects for new technology, such as the testing of a spacecraft or submarine, which would otherwise be life-threatening for a human test subject. Other uses lie in the medical field, helping with surgeries that are long and tiresome for people. Or perhaps, with regard to limiting the spread of disease, humanoid robots can look after patients with highly contagious diseases, such as Covid-19. This allows doctors to have higher attendance, as well as relieving them from the strenuous lives of typically understaffed doctors. These robots will also allow them to focus on more important tasks. One may wonder, “but why can’t you use an ordinary robot for these tasks? Why must they be humanoid?”. Studies show that collecting patient datas and willingness to work with robots is easier when a patient sees some similarity between themselves and the robot. This aligns with the original hypothesis that humans find more humanlike robots attractive. Hence, the need for humanoid robots is evident.
So, how can the uncanny valley be overcome? Is it even possible? Naturally, predicting the human reaction to future robots is extremely difficult, but scientists are highly motivated to try. In the media industry, lifelike people have been created – a term known as “deep-fakes” and are frequently used. Even their voices, put together by computers are unmistakably “human”. This technology can be adapted even further to become the voices of humanoid robots, with natural responses and intonation. Other ways in which engineers are attempting to overcome the Uncanny Valley have to do with the way a robot moves. Programmers aim to mimic every facial expression that a human can do in a natural way. Of course, this is difficult, but studies have shown that a robot with at least 62 facial expressions is seen as more appealing, which has been modelled by some robots in the past, such as Sophia, a robot developed in Hong-Kong. This is clearly successful, as she was voted the most popular speaker at the United Nations Social and Economic Council in 2017. One of the arguably most difficult tasks ahead is that of human personality. The goal is that a robot can have a conversation with a person, without any oddities. Dialogue should be smooth and ordinary, and match that of the other speakers, so that a robot blends seamlessly into society. Programmers do this by creating algorithms for possible scenarios, and creating a robot that learns from its mistakes. This way, a programmer does not have to sit by a computer and control the dialogue, but a robot can deduce the best response to someone by editing previously run algorithms so that they apply better. Extending this, humanoid robots can develop personalities, with catch phrases and likes or dislikes. This seems impossible, but actually, it is already highly integrated into society. For example, Amazon’s Alexa does this very well. As such, all the pieces for building a robot that is indistinguishable from a person are there, they just need to be perfected. With this, the end of the Uncanny Valley is in sight.