In 1966, Joseph Weizenbaum, a computer scientist at the Massachusetts Institute of Technology, made what would today be called a chatbot. Named Eliza, after the fictional Eliza Doolittle from George Bernard Shaw’s Pygmalion, Weizenbaum’s creation was a programme that simulated a therapist. Weizenbaum was less interested in people’s mental health and more in showing how clever programming could make computers — sans context and ‘intelligence’ — appear to have conversations with people.
Sample a conversation, with a typical Eliza programme:
Eliza: Please tell me what’s been bothering you.
You: Work
Eliza: I’m not sure I understand you fully.
You: i can no longer concentrate
Eliza: Can you elaborate on that ?
You: i get distracted too easily
Eliza: Do you say you get distracted too easily for some special reason ?
You: great office coffee
Eliza: Please go on….
And so on..
Eliza is designed to go on as long as power runs out or you’re bored and will diligently return a sentence that is either affirmative or interrogative. No more no less.
Soon Weizenbaum realised that people were hooked on to his creation and would spend hours, even though fully aware that they were talking to a robot that never once blazed the remotest speck of insight.
Eliza spawned many upgrades, research projects and academic research in the domain of computer science but no headway in the field of cognitive or behavioural therapy. Ironicallly, the Higginsian Weizenbaum became a critic of artificial intelligence technology and argued that too much power to machines boded ominously for people.
Circa 2017 , Bengaluru, Jo Aggarwal and Ramakant Vempati, IIT-educated engineers who have decades-long careers in consulting and management ranging from London to Iraq, realised the Eliza could be a Do-A-Lot. With the advances in AI-research and the connectivity made possible by the Web, the duo (who are also married to each other) reckon that Eliza-like bots could be focussed to help people take better care of their moods and tackle depression.
Customised app
‘Wysa,’ an app co-developed by them that has a nod to Eliza in its name, can be downloaded from the Google Play Store. Unlike Eliza, who isn’t committed to a user beyond spitting out single-phrase replies, Wysa is programmed to have customised and an elaborate tree of responses to questions ranging from “I’m feeling low today” to the critical “I feel like killing myself.”
To the former, Wysa prompts you on a conversation-course that includes asking you to “take a deep breath” or suggests some easy-to-do muscle stretching and then prods you to type out what’s precisely upsetting you. In extreme cases like the latter, Wysa can’t differentiate between those actually contemplating suicide or just being hyperbolic but will instantly remind them that they are talking to a bot and suggest a slew of counsellors or therapists they should get in touch with. It will, the next day and unprompted, ask you if you’re still feeling as miserable.
“We cannot prescribe a course of action or replace the therapist, “ says Mr. Vempati. Still in the beta phase, Wysa has had over 30,000 downloads since its launch three months ago and 70% of its users outside of India. Amulya Singh, a 23-year-old engineer, who was recommended by his therapist to try the app, says that he’s been using it for over two months and it has helped him relax. “I like it. I lost my father recently and was depressed and when I can’t reach my counsellor (at night or not scheduled) and feel anxious, I do (use it and) feel I’m talking to a person,” he tells The Hindu.
Discreet
Wysa is one among a slew of bots and applications that Touchkin, the company co-founded by Mr. Vempati and mS. Aggarwal, makes. While Wysa is explicitly tuned to make people feel better and — though buffered by a team of 15 including psychologists and programmers parsing the nuances of speech — Touchkin doesn’t aim to ‘predict’ depression. The self-funded startup aims to focus on those with conditions that aren’t directly connected to mental health. Citing stats that 30% of people with diabetes, and 40-50% of those suffering from cancer or Parkinson’s, are depressed, Ms. Aggarwal says that the company has collaborated with institutions such as Apollo, the Indian Institute of Public Health and insurance companies to measure the extent of improved benefits that Wysa brings. Users, says Ms. Aggarwal, frequently share intensely personal experiences and given that the product is accessible via Facebook opens up concerns of privacy. Mr. Vempati says that people don’t require to share their name, email and contacts to use the app. While it’s possible to glean user location it’s impossible to identify people. “There are ways to make money through user engagement (having users consistently use the product) and also use the data to design public health problems,” says Ms. Aggarwal, “Anyone wanting just the data isn’t someone we wouldn’t engage with.”
jacob.koshy@thehindu.co.in
