Imagine you have an appointment with your therapist; you enter the consulting room, you look up, and look right into two bright glowing robot eyes. How likely is it that a robot will be the future therapist?
ELIZA, your first therapist chatbot
The first therapist chatbot named ELIZA is actually already fifty years old. She was created in the 1960s by Joseph Weizenbaum at the MIT’S Artificial Intelligence Lab. ELIZA is a primitive natural language system that uses pattern matching to parody a Rogation psychotherapist whose main method is to make a patient feel relieved by sharing their problems with a listener. ELIZA is not able to think as a human being, she simply uses linguistic tricks to keep the conversation initiative needed to ensure that the patient is the one who is answering questions and explaining or extending their previous statement. For instance, the word ‘sad’ prompts the reply ‘I am sorry to hear you are sad’. (Rzepka & Araki, 2015) Some people would believe they were talking to a real empathic person. In spite of that, giving advice to a person suffering from mental conditions can go wrong, and the task becomes a moral dilemma.
The next ELIZAs
Since ELIZA, there have been a lot of improvements in computer therapy. There are virtual human agents now that aim to initiate user engagement, increase awareness of symptoms and treatments for users who may be reluctant to talk to a traditional counsellor. (Cameron et al., 2017) Virtual help agents are even counseling Syrian refugees. Can these virtual human agents improve on human therapists? According to Alison Darcy (2017) there is a lot of noise in human relationships. Noise is the fear of being judged. Would people share more if they could talk to an anonymous algorithm?
The moral dilemma
Chatbots can create another option for users who do not want to receive face-to-face treatment, however there are many ethical aspects to consider. Building an empathetic chatbot is a risky place, what if the chatbot provides incorrect or apathetic responses? In my opinion, we should not risk the mental health of human beings yet. For now chatbots can be used to cure loneliness or just as a listener. However, maybe in the future, we will look in those bright glowing robot eyes!
References:
Cameron, G., Cameron, D., Megaw, G., Bond, R., Mulvenna, M., O’Neil, S., Armour, C. and McTear, M., (2017). Towards a chatbot for digital counselling.
Molteni, M. (2017). The Chatbot Therapist Will See You Now. [online] WIRED. Available at: https://www.wired.com/2017/06/facebook-messenger-woebot-chatbot-therapist/ [Accessed 24 Sep. 2017].
Romeo, N., (2017). The Chatbot Will See You Now. [online] The New Yorker. Available at: https://www.newyorker.com/tech/elements/the-chatbot-will-see-you-now [Accessed 24 Sep. 2017].
Rzepka, R. and Araki, K., (2015). ELIZA fifty years later: An automatic therapist using bottom-up and top-down approaches. In Machine Medical Ethics (pp. 257-272). Springer International Publishing.
Dear Sharon,
I think you picked an interesting subject here because initially you would assume that therapists cannot be replaced by Artificial Intelligence. You focussed on the empathic capabilities of a therapist robot and concluded that it is not likely that (in the near future) a robot will be at the same level as a trained therapist.
I would like to look at a robots potential to diagnose mental problems. Through non-verbal and implicit verbal communication we tell much more than we intend to. A therapist is trained to pick up those signals and act upon them, however a therapist is prone to make human mistakes. Using the same techniques used in voice and face recognition, AI could be used in therapy to look for signals of depression, anxiety and other mental diseases.
In my eyes such a robot should be used in combination with a human therapist since, in my eyes, this line of work requires human interaction. The robot could, however, catch diseases much more efficiently and in an earlier stage in order to provide the best help to the patient.
Following this line of thought, machine learning could be used to observe the human therapist at work and perhaps to replace therapists all together, but I doubt that will happen any time soon.
I liked that you brought up that these techniques are used to help Syrian refugees and I agree that many charity organisations could use them. When providing aid in remote areas of the world or after a natural disaster you may encounter many traumatised people who need help fast. Language can be a great barrier in these situations that can easily be overcome by ‘robot therapists’.
In conclusion, I do agree with you that it is (initially) hard for robots to copy and respond to human therapy. However I think these machines can be very valuable in other ways in therapy sessions and provide a quicker and more accurate solution for patients.
Hi Imre!
Thank you for your thoughts on this topic. I love your idea about merging the skills of a robot and a human therapist. This would be a great solution for human flaws (for now). I also agree on helping the Syrian refugees, in times of lack of human therapist that can be a listening ear for these people, it is great that we have chatbots.
Thanks for this inspiring post. I think this topic is very actual and will be applied in the coming next years more and more. You finish your post with your thoughts about applying robotics in the medical health sector. I would like to share my thoughts, because I think robotics should be used in the medical health sector personally.
In the medical health sector, we see that robotics are increasingly used, for example for the diagnosis of some types of cancer or other diseases. The great difference between robotics and human is that robotics do not have real emotions and are more objective. They communicate by following an algorithm and are therefore more able to ‘think’ in a rational and objective way.
For example, IBM developed an application for making decisions when lung cancer has been diagnosed. This application is already used at Memorial Slaon-Kettering Cancer Center. According to IBM, 90% of nurses follow the guidance of this robotic. Don’t you think that’s amazing?
Robotics will thus not only be used for diagnosis but also for treatment. Indeed, for mental diseases this might be more difficult, but in fact it’s also a biological process in the brain cells.
I would like to share the following video with you:
https://www.youtube.com/watch?v=Y9xyNdpbr94
In this video, Michael Mikulsi is telling you more about robotics and biomedical engineering. He designs robots in order to transform phsyiotherapy and healthcare.
Good luck!
Daan Verpalen
Hi Daan! Interesting video and thoughts, thank you. I agree on the fact that robots can be very useful in the health sector. On the other hand, mental diseases are very hard to explain in an algorithm. Maybe if we can sort out the biological process in our brain cells and turn them into algorithms, it would be a major change!
Nice article Sharon! Thanks for sharing.
I think that everyone has a different level of comfort with robots. The Japanese, for example, are interacting a lot more with robots via arcade games, entertainment dinners and devices/applications such as ‘ Waifu / wife’ from Gatebox. She wakes you up in the morning, gives you the weather forecast, turns on your heater and lights etc. For some people, robots would be very helpful, whereas others will prefer not to use them.
I agree that robots miss the necessary empathy that real humans have and I am skeptical they will ever learn this, so if Robots are ever to be used in the mental health industry, they should stick to diagnosing and leave the real treatment to humans.
Hi Inge, interesting thoughts! It is also very important to take into account different cultures, if we want to implement human therapists. Every culture reacts differently on these kind of technologies. Maybe we should start in Japan, since they are already familiar with robots in their life!
This is a very interesting dilemma. As Inge Schenk has also mentioned, our view on robots is also determined by how open we are for them (through our culture).
At the moment, robots indeed miss the empathy and emotional intelligence that we humans have. However, AI is getting really good at pinpointing and diagnosing mental health issues. One example is this research, where AI diagnosed predictive markers of depression from your instagram feed.
Coming back to the empathy and emotional intelligence of AI, there are signs that chatbots are becoming better at providing online customer experiences as they do not have “bad days” or a dislike of annoying customers.
And also, sometimes it is easier to talk to non-humans about mental health issues, as people are prone to omit or lie to their therapist, as they are afraid that they will be judged by them (possibly due to social anxiety). However, by interacting with non humans, robots, some patients will benefit greatly.
There are many pros and cons on this topic. I am very excited, though, what the future will bring!
Links:
– https://epjdatascience.springeropen.com/articles/10.1140/epjds/s13688-017-0110-z?sf104446268=1
– https://www.forbes.com/sites/shephyken/2017/07/15/ai-and-chatbots-are-transforming-the-customer-experience/#7520d28641f7
Thank you for your thoughts! Interesting example of the Instagram feed diagnose. That is also a way to identify people that suffer from mental diseases. On the other hand, not everyone who suffers from these diseases want help from others. So not everyone that will be recognized on Instagram wants to accept the help or diagnose that is offered after the diagnosis. A lot of pros and cons on this topic!
Inspiring post Sharon
Inspiring, and very interesting, post!
I was wondering if you have heard about Woebot? Woebot is an artificially intelligent chatbot programmed using Cognitive Behavioral Therapy (CBT) that was designed for people with depression:
https://www.businessinsider.nl/stanford-therapy-chatbot-depression-anxiety-woebot-2017-6/?international=true&r=US
Woebot was designed by Allison Darcy, a clinical psychologist at Stanford, who has tested the efficacy of this chatbot in a clinical study with 70 subjects. The study itself was published in the Journal of Medical Internet Research Mental Health, and is a fascinating read.
Anyway, previously I did not think chattiest could be a real substitute for therapists. I cannot say my opinion has fully changed on this front, but I am at least open-minded to the idea that Woebot could be a strong mental health resource for individuals who cannot afford therapists or may need mental health services sooner than their next appointment with their traditional therapist. If anything, I find Woebot as a real useful tool potentially in depression management to be used IN TANDEM with whatever therapy visits is helpful to the individual since it gives them the ability to chat with a CBT-coded chatbot whenever and wherever, at the individual’s convenience. Indeed, “A review of studies published recently in the journal World Psychiatry compared people who received CBT online with people who received it in person and found that the online setting was just as effective.” Perhaps chatbots are a real potential way for therapy to be accessible for all, and not just a privilege for the wealthy (as is the case where I am from – the United States).