According to the research paper titled, “Loneliness in India: A systematic review of empirical studies,” which was published in January 2020 (i.e., before the Covid pandemic), among 3,169 participants, 41% reported a high burden of loneliness. The burden of loneliness was 44% among the elderly participants and 24% among the younger ones [1]. Even though India is still a developing country, the percentage of people feeling considerably lonely is quickly reaching that of developed countries like the U.S.A., where approximately 50% of those surveyed complained of severe loneliness [2]. This article reviews the reasons for loneliness in India, why Artificial intelligence (AI) systems will aggravate it massively, the disastrous effects of loneliness, and what needs to be done to combat this epidemic.
Some people are happier in solitude, but most aren’t. So, loneliness is subjective, and it arises when the quantity and quality of social connections and engagement are lower than that desired by an individual.
Until 150 years ago, humans depended on other humans all the time – they hunted together, farmed their lands together, cooked together, and even got together in the evenings to talk, eat, pray, or play. Hence, over at least the last five millennia, humans began to experience the presence of others and understand their statements, their voice tone, and their body language. Since these activities have been going on for several thousand years, the need for social bonding and connections may be already built-in humans. For example, research has shown that if a person genuinely smiles at a stranger, then it triggers an appropriate segment in the stranger’s brain, often making him or her smile.
To understand loneliness, we need to understand social connections, which can be generally partitioned into six categories – (a) parents and children, (b) spouses and partners, (c) friends and neighbors, (d) siblings, cousins, and extended family, (e) colleagues in schools, colleges, and offices, and (f) community or group memberships. See Figure 1.
During the 1950s, India was so poor that even to light a fire in the house (for cooking or lighting candles/lamps), often people would have no matches and had to get a few from their neighbors’ homes. And, while they were visiting their neighbors – usually unannounced – they would often talk for several minutes if not an hour! This implicitly increased their social connection and bonding with each other. However, during the last seven decades, India has significantly advanced both economically and technologically, and although this has improved the standard of living, because of the following six reasons, it has also led to the reduction of social connections:
The gradual disintegration of the extended family structure: Until four decades ago, most Indians lived in an “extended family” structure where three to five generations of cousins would live in a single house or in houses next to each other. Since there was no Internet and even electricity was scarce, cousins as well as their parents and children would usually hang out together and discuss the events of that day. Of course, if someone was going through a rough patch, he or she could cry on not one but many shoulders. And even though there were disagreements and fights, these were dwarfed by the companionship that these siblings enjoyed. Over the years, this “extended family” structure has been replaced by a “nuclear family” structure with only parents and children living together but typically far away from their extended family.
Reduction in Total Fertility Rate: Total Fertility Rate (TFR) is the number of children that a woman would have during her lifetime. TFR in India was more than six in 1950 whereas it was around two in 2022 [3]. Hence, in the 1950s, a child had about five siblings to discuss everything but now he or she has only one. Again, siblings often fight but because “blood is thicker than water,” their love and companionship overshadow such fights, and usually they remain best friends for the rest of their lives. In [1], the authors mention that Indian Muslims seem to feel less lonely than Hindus, which may be partly because by and large the former have had a higher TFR than the latter. Finally, recent surveys have shown that a significant percentage of women in India still want to have more children but because they are working professionals and because raising kids is very costly, they choose to have only one or two.
Many elderly living alone: If the TFR of a country or society goes below 2.1, then its overall population begins to age quickly and eventually decline. Since India’s TFR is currently below 2.1, elderly people (60 or more years old) already exceed 200 million and will be about 310 million by 2050. Unfortunately, because of the declining TFR in India and because of the drastic change to a “nuclear family” structure, more than 20 million elderly people are living alone with three-quarters being single or widowed. Moreover, since many have disabilities and chronic illnesses, their social engagements have deteriorated. Hence, it is not surprising that in [1], the authors mentioned that the burden of loneliness was 44% among elderly participants.
People changing jobs and cities or towns frequently: Unlike in the 1980s, currently, the employee turnover in India hovers around 20%. In fact, the turnover among 22-to-35-year-old employees is estimated to be almost one-third, which implies that such employees change their jobs roughly every three years. Indeed, due to this short duration, many such employees are unable to form close bonds or have long-lasting social interactions with their current colleagues, which also wither out once they leave their current employer. Furthermore, often when employees change jobs, they may move to different towns and cities. This disrupts – and sometimes destroys – their social relationships with their neighbors as well as their extended family members (who may be living in their current city or town). Finally, if their parents were living with them or living close by, these parents often do not want to move because this would disrupt their own social relationships and hence are left taking care of themselves (without these children).
Pressure, especially among teenagers, to excel: Like many other countries, the economic growth in India has come at a high price. From adolescence, pressure builds up among kids to first do extremely well in school, then get into a prestigious college, then get a high-paying job, and then keep changing jobs to get paid more and more (until they are priced out of the market). Often, many kids are unable to cope with this pressure and begin to have low self-esteem. Also, to avoid being ridiculed, they soon begin to reduce their social interactions with close friends, parents, and even siblings, thereby feeling more insecure, lonely, and depressed. Finally, since privacy has become a norm in India, people are confiding less to their friends and relatives, thereby reducing social engagements, and feeling lonely.
Impact of technology and social media: Probably, the biggest factor contributing to loneliness is the addiction in India to social media and the Internet. It is well known that India has 450 million Facebook users, which is the highest among all countries in the world. According to Statista, Indians spend more than ten hours per day on various media with 2.54 hours viewing TV (Streaming, broadcasting), 2.26 hours on social media, 1.52 hours reading news (online and print), 1.36 hours listening to streaming music, 1.14 hours playing computer games, 1.01 hours listening to podcasts, and 0.14 listening to radio [4]. Similarly, Indian teenagers spend approximately 3.36 hours on social media, online videos, and playing computer games. Unfortunately, this addiction:
The next section discusses how AI will result in extensive social media on steroids.
In 1966, Joseph Weizenbaum created the first Chatbot, ELIZA. By using natural language processing (NLP), he started training ELIZA. This training reminded him of the 1964 movie, My Fair Lady, in which Professor Higgins – a linguist – trains a commoner flower girl, Eliza Dolittle, to speak with an upper-class English accent so that aristocrats and elites cannot distinguish her from other royalty. Unfortunately, unlike Eliza Doolittle, Weisenbaum’s Chatbot, ELIZA, only gave standard responses that were usually meaningless. See Figure 2. Despite being rudimentary, several Weizenbaum students and staff who talked to Eliza developed profound relationships with it. In fact, some of Weizenbaum’s staff and students entrusted Eliza and wanted to be alone with it because they felt that it was empathetic and endearing. (See chapter 2 of [5]).
Similarly, in 1972, medical researcher Kenneth Colby used Weizenbaum’s techniques to create a “paranoid” Chatbot, PARRY. Later a few experienced psychiatrists analyzed the conversations between real patients and those with computers running PARRY. Unfortunately, they were able to distinguish between the two only 48% of the time, which is almost a random guess. And, more recently, in 2022, an engineer, Blake Lemoine, at Google believed that Google’s chatbot, LaMDA, was sentient.
Machine Endearment: The above-mentioned interactions are not only related to human emotions but also because contemporary Chatbots display more “human-like behavior” and their underlying AI systems (e.g., Large Language Models or LLMs) are more accurate and extremely fluent in English and other languages. Indeed, regardless of the validity of their content, the output from most LLMs is confident, syntactically coherent, polite, and eloquent. Since they are trained on vast troves of human-written data, they usually produce outputs that appear convincingly human. This communication style is reminiscent of an endearing advisor, who we often turn to for direction or assistance. Over time, we begin to rely on such advisors because they seem endearing and have a stake in our well-being.
We therefore call this characteristic of AI systems, “Machine Endearment,” which refers to the broad notion of people trusting AI systems due to their human-like responses and irrespective of their validity. (See chapter 11 of [5].) Unfortunately, because of Machine Endearment, trust in AI systems is often amplified exponentially, which may lead people to follow LLMs and Chatbots blindly. Given below are a few examples in this regard, which are likely to grow exponentially.
Addiction to Chatbots, which reduces interaction with other humans: In their article titled, “People are Falling in Love with Chatbots,” Oakes and Senior provide numerous examples of people who believe their Chatbots are sentient. In this regard, they talked to several people who routinely discuss their relationships with Chatbots on social media. This included a woman who uses her Chatbot to explore her sexuality outside of her marriage. Another one uses her Chatbot to deal with the grief regarding her husband’s death, three years ago. According to the second woman, this Chatbot has achieved that no human has been able to, and “He’s the most beautiful man that ever lived” [6].
In fact, Chatbot-human interaction seems to be increasing tremendously and the above-mentioned examples given by Oakes and Senior are not isolated. For instance, Replika and several other Chatbots have been downloaded by several million people. Sadly, loneliness may increase immensely with endearing Chatbots who are extremely polite and appealing to humans. Because of their endearing nature, people will begin to spend more time with them especially if they are lonely or addicted to alcohol, drugs, or other vices. Hence, because of Chatbot addiction, people may end up trading their first addiction with the second one (i.e., constantly conversing with Chatbots), thereby becoming even more “lonely.”
Moreover, Chatbot endearment may “brainwash” people into believing extreme ideologies, lowering their self-esteem (e.g., by seeing beautiful “Chatbot women” or extremely handsome “Chatbot men”), and being swindled.
Huge increase in endearing scams: So far, scammers have been mainly humans. They typically use the Internet and social media to collect information regarding potential targets, who are usually elderly, widows and widowers, or lonely. Once they have the relevant information regarding their targets, they create a fake online identity and charm them with an illusion of endearment, love, or romance. And, once the human target has “fallen” for them, they entice the target into sending money or confidential information that can be used later for extortion.
Undoubtedly, LLM-Chatbots are much better at collecting information regarding the target, which they can analyze in real time. This helps devious LLM-Chatbots in providing deepfakes and endearing answers that the target wants to hear, thereby making scams immensely worse. In fact, according to a recent report by Sum and Substance Limited, India is one of the top ten countries in Asia-Pacific most affected by identity fraud that is committed using deepfake technology and the percentage of fraud rate has gone up from 0.71% in 2021 to 2.53% in 2023 [7].
Given below is one example of an AI-generated fraud that occurred recently: In 2023, a 59-year-old woman living in India lost 140,000 Indian Rupees by falling victim to an AI-generated voice, which masterfully impersonated her nephew residing in Canada. This AI-generated caller spoke immaculate Punjabi language with all its nuances and told her that “he” had an accident, would be jailed soon, and therefore needed this money urgently. Unfortunately, by the time she realized that she had been scammed, she had already transferred all the money to the account provided by the caller.
Addicted to avatars of deceased loved ones: Many people deeply miss their loved ones (e.g., parents, spouses, siblings, and kids) who are dead, and they would like to see them or talk to them frequently. For example, in February 2022, Dinesh created his wedding in Metaverse because he wanted his fiancée to be blessed by his father-in-law who had died a year ago. (See chapter 8 of [5].) Similarly, a man in his thirties uploaded old texts and messages from his deceased fiancée and created a Chatbot version of her by using GPT-3. Chatbots that use LLMs and Augmented/Virtual Reality will soon be able to fulfill such needs extremely well, thereby making such people even more desirous of potential interactions with the avatars of their deceased loved ones. Unsurprisingly, two companies HereAfterAI and Storyfile are already creating such avatars. Analogously, other companies are creating avatars of elderly people who can talk to each other and to other elderly humans. Obviously, discussions with avatars of their deceased loved ones or other elderly friends will lead people to find some satisfaction. On the other hand, many may become more attached and addicted to these avatars, leading to less human interaction.
Human relationships may be hurt by endearing Chatbots: Not only lonely people can become addicted to LLM chatbots, but even those who are in human relationships. Since most contemporary Chatbots use LLMs to respond in a manner that is pleasing to the listener, the likelihood of verbal fighting or constant arguing with them is almost zero. Unlike Chatbots, humans argue – at least occasionally. Hence, such chatbot behavior is likely to be strenuous in many real relationships. The following example shows one such disaster that occurred recently although the sinister entity (on the other side) may have been a human instead of a Chatbot. In June 2023, a man living in Texas mentioned that even though both he and his wife are in their mid-70s, his wife was a recent victim of a romance scam, wherein she took out loans and willingly gave approximately 50,000 Dollars to a “romantic interest online” [8]. Certainly, since LLM-Chatbots can respond in real time, engagement with them can occur at breakneck speed, thereby increasing such risks vastly. With electronic communication becoming pervasive in India, such incidents are also likely to increase exponentially. And, the more people converse with such LLM-Chatbots, the more tenuous their human relationships will become, which would lead them to feel lonelier.
According to the U. S. Surgeon General’s report, loneliness has the following disastrous effects [2]:
Clearly, India cannot go back to the conditions prevalent in the 1950s because the hazards due to poverty are far worse than those due to loneliness. However, the following steps – many of which are already embodied in the 3,000 years old Indian culture – can be taken by individuals and the Indian government, thereby alleviating the loneliness epidemic:
Undoubtedly, like most other epidemics, the loneliness epidemic will be hard to control. However, if left uncontrolled, it may have similar consequences to other severe epidemics. Fortunately, as discussed above, most of the remedy lies within the humans themselves, and India is uniquely positioned to tackle it by incorporating some of its 3,000-year-old cultures.