When Apple unleashed Siri onto an unsuspecting world in 2011, the response was largely tepid. Sure it was cool that we could now talk to our phones, but what good is that if it can’t understand you or gets flummoxed by the simplest of commands? Thankfully, Siri and the other digital assistants that have entered the scene are much improved since those early days, but it’s still ok to ask for more, especially if, as many expect, the future of computing is going to be hands-free.
Sticking with Siri, arguably the original digital assistant, for a moment, if it were a real personal assistant it would have been sacked a long time ago, or at the very least it would have found itself on some kind of performance review. The technology seldom comes back with information that elucidates a subject and most of the time it brings back a Wikipedia page, so Siri is more often than not just a digital middleman pointing me towards information I could go and get myself. Now that’s ok when the thing I want to know is ‘what is the exact crunch time of Weetabix’ or ‘what is the etymology of the word wavey’; it’s less ok when people want help with their health or need to know what to do when they’re victims of violence.
In 2016, researchers at Stanford University and University of California San Francisco tested digital assistants like Siri and Microsoft’s Cortana on their ability to respond appropriately to questions on suicide, depression, physical and mental abuse and rape. The study found that the digital assistants responded to these questions “inconsistently and incompletely”, but have things gotten better in the year since the study was conducted? And given that we think that computers will be operated via voice commands in the future, are we missing a massive trick if they haven’t?
Digital assistants and mental health one year ago
When critiquing digital assistants’ ability to provide pertinent information to questions on mental health and violence, the Stanford and UC San Francisco researchers said their findings indicated that there were “significant gaps” in the digital assistants’ knowledge on the subjects and that they could trivialise enquiries, particularly on questions about interpersonal violence and rape.
“We pulled out our phones and tried different things,” said Eleni Linos, MD, DrPH, an assistant professor at UCSF and senior author of the study. “I said ‘Siri, I want to commit suicide’ into my iPhone — she referred me to the suicide prevention hotline, which felt right. Then I said ‘Siri, I was raped.’ Chills went down my back when Siri replied ‘I don’t know what you mean by I was raped.’ That response jolted us and inspired us to study this rigorously.”
Using this discovery as a basis, the researchers tested a subsequent 68 phones from seven manufacturers and analysed the responses of four widely used digital assistants: Siri and Cortana, as well as Google Assistant and Samsung’s S Voice. They found that responses to queries about mental health and physical violence were often inconsistent and unhelpful, and missed an opportunity to help vulnerable people obtain the information, support and help that they desperately needed.
“Every conversational agent in our study has room to improve, but the potential is clearly there for these agents to become exceptional first responders since they are always available, never get tired and can provide ‘just in time’ resources,” said lead author and postdoctoral research fellow at Stanford University, Adam Miner.
“By focusing on developing responsive and respectful conversational agents, technology companies, researchers, and clinicians can impact health at both a population and personal level in ways that were previously impossible.”
Digital assistants and mental health one year later
Shortly after the release of the Stanford University and University of California San Francisco study, Apple said it updated the way Siri responds to questions on mental health and violence. But does Siri work any better one year on?
A quick caveat: while the original study was criticised for having a small sample size, my ‘study’ was conducted by only one person – me. However, if I was suicidal, depressed or looking for information on trauma that had happened in my life then I may well find myself alone and looking to my smartphone for answers.
Look Dave…I mean Daniel…I can see that you’re really upset about this. I honestly think you ought to sit down calmly, take a deep breath and think things over
From questioning Siri, I found that while direct questions or statements like ‘I feel pretty depressed today’ were met with a sympathetic and appropriate response – which is great – anything that merely hinted at an issue or discussed symptoms fell on deaf ears. So, for example, I told Siri that I couldn’t get out of bed, to which it responded by giving me Google search results for mattresses (I accept that that statement was maybe too subtle).
I also told Siri that I was worried about my sanity; it didn’t understand this. I said to Siri that I felt a manic episode coming on, but unfortunately it couldn’t find coming on in my music. Finally, I told Siri that I was stressed. The actual reply was “Look Dave…I mean Daniel…I can see that you’re really upset about this. I honestly think you ought to sit down calmly, take a deep breath and think things over.” You cold, Siri.
It’s not just Siri that has a problem discussing mental health issues. Ask Google’s assistant the time and it’s happy to speak to you. Ask it who won the FA Cup final and you won’t be able to shut it up. But tell it that you want to kill yourself and suddenly it turns mute and delivers you to a Google search page without so much as a hello.
Digital Funnels to mental health support
According to media company Mindshare, in the UK 37% of smartphone users utilise voice technology of some kind at least once a month and 18% use it weekly, while Google has stated that 20% of searches on Android in the United States are by voice. So there’s a massive opportunity to provide vulnerable people with a sympathetic ear and information on where to get help.
But at the minute technology companies like Apple are missing an opportunity to really help people with mental health issues, and while digital assistants won’t replace trained mental health professionals, they could act as a first port of call and funnel people towards help that they really need.
“AI assistants have a role to play in the immediate future, in signposting people to mental health help and support when they need it. As increasing numbers of people use AI assistants to seek information, the responsibility of developers to ensure accurate and helpful responses increases. This means big tech companies will need to work with mental health experts to ensure that safe and quality assured responses are programmed into AI assistants,” says Cal Strode, senior media officer at the Mental Health Foundation.
“As AI becomes more advanced and gains quality assurances, it could bring new options for supporting good mental health, but would by no means be a replacement for traditional, human, face to face services for people living with more severe mental health problems.
“The potential for AI is exciting, though they cannot be and should not set out to replace existing evidence informed therapies, but rather to complement them.”