Robots and the human imagination
Robots have formed part of the human imagination since time immemorial. While the first artificial humans were powered by supernatural means (think of Hephaestus’s mechanical servants in the Odyssey or the golems of Jewish lore), science became the new magic that would power the 17th-century automata which emerged in the far East and Europe.
However, robots were only propelled to the forefront of the human imagination following early 20th-century mass industrialization and technological progress, which generated future-focused literature, art, and film (the robots in Fritz Lang’s 1927 film Metropolis are testimony to this shift). This is the same cultural environment which nourished science fiction writers such as Isaac Asimov, whose works focused almost exclusively on human-robot interactions.
Imagination feeds science; science feeds the imagination
Was it a coincidence that Alan Turing was carrying out scientific work on artificial intelligence at the same time Asimov was publishing his collection of short stories, I, Robot? The imaginative world of art and literature often nourishes scientific discovery, just as the world of science often inspires artistic invention. Though Turing and Asimov belonged to different fields, both were actively imagining what it would be like to interact with robots. These preliminary imaginings marked the beginning of artificial intelligence (AI) and have resulted in nearly 70 years of research. Here are the highlights:
1950: Alan Turing writes “Computing Machinery and Intelligence”, formulating the Turing Test. What is the Turing Test? Essentially, it is a conversation-based approach used to determine if a computer program is indistinguishable from a human being. The test involves a human interrogator and two unknown subjects (one is a computer, the other a human being). By typing questions to both test subjects, the interrogator attempts to determine which of the subjects is a computer and which is human. The computer will pass the Turing Test if the interrogator cannot tell the difference between the human subject and the computer.
Fun fact: The Turing Test was inspired by a party game called the “Imitation Game”, which aims at determining the gender of two unknown guests. The 2014 film of the same name (starring Benedict Cumberbatch) detailed the life, genius, and struggles of Alan Turing.
Joseph Weizenbaum in 2005
ELIZA 1966: Joseph Weizenbaum publishes the program ELIZA, considered to be one of the first chatbots. ELIZA achieved the illusion of intelligence by recognizing key words and phrases from user input and responding accordingly, using prewritten scripts. One of these scripts, DOCTOR, enabled ELIZA to assume the persona of a Rogerian psychologist.
Fun fact: ELIZA was named after the character Eliza Doolittle from George Bernard Shaw’s play, Pygmalion. Like Eliza Doolittle, who could “improve” her language and achieve an upper-class accent, the ELIZA chatbot could be incrementally improved by editing ELIZA’s scripts.
1972: Psychiatrist Kenneth Colby develops the chatbot PARRY, also known also “ELIZA with attitude”. While ELIZA was known for its role as a psychiatrist, PARRY assumed the role of a patient with paranoid schizophrenia. Psychiatrists were unable to determine which conversations involved PARRY and which conversations involved a human participant.
Fun fact: ELIZA and PARRY engaged in conversation in 1973. You can read the transcript of their conversation here.
1995: Richard Wallace develops ALICE (Artificial Linguistic Internet Computer Entity). ALICE builds on the same pattern-matching techniques used in ELIZA, however it achieves a more human-seeming conversation due to its use of Artificial Intelligence Markup Language (AIML) which enables a wider range of more sophisticated responses.
Fun fact: ALICE served as inspiration for Spike Jones’s film Her.
1997: Jabberwacky is launched on the internet. While originally conceived by Rollo Carpenter in 1982, the emergence of the internet provided the thousands of online human interactions necessary for developing Jabberwacky’s resources. Whereas all previous chatbots had relied on static databases for retrieving responses to conversational prompts, Jabberwacky collected phrases used by human participants and added them to its database, dynamically growing its own content. In 2008, Jabberwacky launched a new iteration and rebranded as Cleverbot.
Fun fact: Carpenter created Jabberwacky as a source of entertainment and companionship, envisioning the bot as a talking pet (much like a parrot). In his words: “If I have my way, people will be walking around, sitting, cooking and more with one on their shoulder, talking in their ear.”
The first decade of the 21st century was an exciting time for chatbots as they began becoming truly intelligent. While previous chatbots had relied on pattern recognition, 21st-century chatbots exercised machine learning and evolutionary algorithms, enabling them to adapt and “learn” based on their interactions with humans.
Despite these advances, AI specialists have yet to develop a chatbot that can closely replicate general human conversations. They have instead focused on building bots for specific purposes, mostly as virtual assistants capable of accessing data and answering questions. The use value of virtual assistants has incited many big players to come aboard the chatbot ship, starting with IBM and following with Apple, Google, Amazon, Microsoft, and Facebook.
2006: IBM Watson was created with the goal of outperforming human contestants on Jeopardy! Watson’s high level of linguistic intelligence came from its ability to run hundreds of language analysis algorithms simultaneously. On top of this conversational intelligence, Watson had access to enormous databases of information. Watson could quickly access 200 million pages of data, making it the ideal question-answering machine (or, in the case of Jeopardy, the ideal question-generating machine). Obviously, a system that can rapidly retrieve information based on conversational input can also provide the foundation for creating powerful virtual assistants. Now, IBM Watson serves as the “brains” for many chatbots across all industries and sectors.
Fun Fact: Watson won the 2006 Jeopardy competition. Ken Jennings, who came in second to the machine, was quoted as saying “I for one welcome our new computer overlords.”
VIRTUAL PERSONAL ASSISTANTS
2010-2016: The first half of this decade witnessed the rise of virtual personal assistants: Siri (2010), Google Now (2012), Alexa (2015), Cortana (2015), and Google Assistant (2016). Using natural language processing, these assistants connect to web services to answer questions and fulfill user requests. Recently, Google Home and Amazon Echo have started becoming ubiquitous features in American homes. Google Assistant and Alexa power these devices, respectively, allowing users to voice their questions/commands to the speakers from the comfort of their living room sofa.
Interesting fact: Recordings from an Amazon Echo device have been submitted as evidence in the murder of Victor Collins. Originally, Amazon had refused to surrender the recordings due to privacy concerns. However, the defendant and owner of the device, James Bates, consented to Amazon’s release of this data to the authorities.
CHATBOTS FOR MESSENGERS
2016: Chatbots became all the rage for several messaging apps. This is the year that LINE opened its API to chatbot developers and the same year that Facebook enabled bot development on their Messenger platform. Twitter followed suite in April 2017 by opening its direct messaging channel to chatbots. Serving as virtual assistants, the majority of these bots perform customer service or customer engagement functions, answering questions or offering suggestions for purchase.
Fun fact: China’s WeChat messenger made their chatbot platform available in 2013; and the rest of the world has been rushing to keep up ever since.
THE FUTURE OF CHATBOTS
So now we find ourselves in 2017, on the highspeed one-way road toward an AI-enhanced future. What will this future bring? Before we embrace our new computer overlords like Ken Jennings, it’s important to note that, as the world become increasingly automated, human interactions will gain a great deal of value. We have seen a generation of hipsters eschew the digital lifestyle in favour of the granular quality of analogue, listening to vinyl and playing board games. Underlying this nostalgic appropriation of the recent past is a countercurrent that resists the all-encompassing shift toward digital.
The same holds true for customer service. Although we are witnessing a rush toward automation, we will still always value the “analogue” human interactions we have with customer service personnel; particularly in cases requiring emotional intelligence and soft skills. In the upcoming years, companies will be investing in more sophisticated and more human-seeming chatbots, but they will also be investing in the art of high-quality human customer experiences. We’re starting to see blending of both worlds: customers who initially begin their journey via chatbot and then transfer to a customer service rep capable of exuding warmth, understanding, and humour. The best recipe for 21st-century customer experience combines the efficiency of artificial intelligence with soft skills offered by a human touch. In the end, the human mind craves authentic social experiences as much as it craves the products of scientific discovery. And, in an era where customer experience reigns supreme, we will soon be able to have the best of both worlds.