Could robots be counselors? Early research shows positive user experience

[Indigophoton]

It appears that therapists need not even be human when it comes to delivering therapies aimed at behavioural change,

New research has shown for the first time that a social robot can deliver a 'helpful' and 'enjoyable' motivational interview (MI) - a counselling technique designed to support behaviour change.

Many participants in the University of Plymouth study praised the 'non-judgemental' nature of the humanoid NAO robot as it delivered its session - with one even saying they preferred it to a human.

Led by the School of Psychology, the study also showed that the robot achieved a fundamental objective of MI as it encouraged participants, who wanted to increase their physical activity, to articulate their goals and dilemmas aloud.

ABSTRACT
Background: Motivational interviewing is an effective intervention for supporting behavior change but traditionally depends on face-to-face dialogue with a human counselor. This study addressed a key challenge for the goal of developing social robotic motivational interviewers: creating an interview protocol, within the constraints of current artificial intelligence, which participants will find engaging and helpful.

Objective: The aim of this study was to explore participants’ qualitative experiences of a motivational interview delivered by a social robot, including their evaluation of usability of the robot during the interaction and its impact on their motivation.

Methods: NAO robots are humanoid, child-sized social robots. We programmed a NAO robot with Choregraphe software to deliver a scripted motivational interview focused on increasing physical activity. The interview was designed to be comprehensible even without an empathetic response from the robot. Robot breathing and face-tracking functions were used to give an impression of attentiveness. A total of 20 participants took part in the robot-delivered motivational interview and evaluated it after 1 week by responding to a series of written open-ended questions. Each participant was left alone to speak aloud with the robot, advancing through a series of questions by tapping the robot’s head sensor. Evaluations were content-analyzed utilizing Boyatzis’ steps: (1) sampling and design, (2) developing themes and codes, and (3) validating and applying the codes.

Results: Themes focused on interaction with the robot, motivation, change in physical activity, and overall evaluation of the intervention. Participants found the instructions clear and the navigation easy to use. Most enjoyed the interaction but also found it was restricted by the lack of individualized response from the robot. Many positively appraised the nonjudgmental aspect of the interview and how it gave space to articulate their motivation for change. Some participants felt that the intervention increased their physical activity levels.

Conclusions: Social robots can achieve a fundamental objective of motivational interviewing, encouraging participants to articulate their goals and dilemmas aloud. Because they are perceived as nonjudgmental, robots may have advantages over more humanoid avatars for delivering virtual support for behavioral change.

J Med Internet Res 2018;20(5):e116
doi:10.2196/jmir.7737

(Emphasis added).

"The next stage is to undertake a quantitative study, where we can measure whether participants felt that the intervention actually increased their activity levels."

A quantitative study that will measure...whether people think they are doing more. Based on the initial study, and what we know about subjective measures of talk therapies for behavior modification, I predict the study will show that human therapists can be exchanged for robot ones with little loss of efficacy

https://techxplore.com/news/2018-05-robots-counselors-early-positive-user.amp

 

[Hoopoe]

No way that robots can replace humans. Unless maybe the robots wear blue cardigans or some similarly impressive equivalent.

 

[Webdog]

On the plus side, all the suddenly unemployed therapists will have "someone" to talk to.

 

[Trish]

"The next stage is to undertake a quantitative study, where we can measure whether participants felt that the intervention actually increased their activity levels."

No, you numpties, the next stage is to do a controlled trial with groups having robot therapist, real therapist, an exercise class and no therapy and all wearing actometers throughout the trial. Or maybe give them free dance classes or gym membership and see if that works better. Or maybe give them a robot to play with as a reward for exercising more. So many possibilities.

 

[NelliePledge]

From my experience of IAPT CBT s robot would probably be as good as the human. No criticism of the person involved it isn’t a process that requires much human input or empathy.

 

[Trish]

It would be better than the CBT I had for feeling miserable about having to give up my job because of ME. At least the robot would not deny my reality and tell me my perceptions of my experience were faulty and all I had to do was feel the opposite of what I was feeling. Or some such twaddle.

Edited for clarity.

 

[Webdog]

Perhaps robots would be better than humans at determining which diagnosed psych patients are actually misdiagnosed ME/CFS patients.

 

[Arnie Pye]

I would assume that any "consultation" with a robot was being recorded, and I would be worried about who was listening to it. If I speak to a human therapist at least I have a real person to liaise with. But with a robot? I wouldn't trust anyone to tell me the truth about who has access to the recording, and it isn't possible to take a robot to court about betraying confidential information.

 

[Invisible Woman]

I would have a serious issue....

I love figuring out how things work and taking them apart. I still remember Cindy-gate when I dismantled my sister's doll because I wanted to know how they got her arms to bend (elastic internally in the joint).

Turns out my ability at reassembling wasn't quite as good.

Could I just play with the robot instead?

 

[arewenearlythereyet]

No, you numpties, the next stage is to do a controlled trial with groups having robot therapist, real therapist, an exercise class and no therapy and all wearing actometers throughout the trial. Or maybe give them free dance classes or gym membership and see if that works better. Or maybe give them a robot to play with as a reward for exercising more. So many possibilities.

Don’t forget the pottery classes

 

[Wonko]

As it's considered unethical for a therapist/counselor and their patient to be overly "friendly" does that mean it's ethically and morally fine (in fact practically a public duty) for a patient to unscrew a robot therapist?

 

[arewenearlythereyet]

 

[alktipping]

the quality of the robots coding would be down to the people who design the code or at least the parameters of the code . which in the case of b p s lot guarantees gigo garbage in garbage out . but nice would give it the go ahead if it meant one robot giving a series of "supportive" classes to all patients newly classified as having somataform disorders . think of the savings thousands of patients being brainwashed into believing all their health problems were caused by faulty beliefs and over awareness of their bodily symptoms. by just one robot or a recording of said robot after all once it has completed the set amount of classes you only have to show the recordings to the next groups of patients.

 

[Forbin]

No way that robots can replace humans. Unless maybe the robots wear blue cardigans or some similarly impressive equivalent.

How about a blue t-shirt? What could go wrong?


"Hear me now and believe me later. You look like a flabby little girly man. Have you seen Sarah Connor? You need more exercise! I'm going to recommend some CBT. That's 'Conan's Barbarian Therapy.' Have you seen Sarah Connor? No, I am NOT a robot! I am a Cyberdine Systems series T-800 Model 101 Graded Exercise Terminator, and I am here to pump you up! Are certain you haven't seen Sarah Connor? OK. I going to prescribe a phased plasma rifle in the 40-watt range. I think I have a sample outside. I'll be back."

 

[Webdog]

"Hear me now and believe me later. You look like a flabby little girly man. Have you seen Sarah Connor? You need more exercise! I'm going to recommend some CBT. That's 'Conan's Barbarian Therapy.' Have you seen Sarah Connor? No, I am NOT a robot! I am a Cyberdine Systems series T-800 Model 101 Graded Exercise Terminator, and I am here to pump you up! Are certain you haven't seen Sarah Connor? OK. I going to prescribe a phased plasma rifle in the 40-watt range. I think I have a sample outside. I'll be back.

"Talk with me if you want to live."

 

[FreeSarah]

Crude stuff. What we need is Artificial Intelligence, not robots.

We programmed a NAO robot with Choregraphe software to deliver a scripted motivational interview focused on increasing physical activity.

i.e. the people programming the robot have already decided what’s desirable and how the task should be performed.

Lead academic Professor Jackie Andrade explained that, because they are perceived as nonjudgmental, robots may have advantages over more humanoid avatars for delivering virtual support for behavioral change.

'perceived as'. The whole value of where we're going with technology now is that is IS non-judgmental, not that it's perceived that way. Programming a 21st century technology system to deliver a human-scripted motivational interview is like solving a complex equation on your laptop by writing the figures and symbols on the screen with a Sharpie pen. It’ll sort of work, but only if your maths is good enough in the first place. The machine would certainly do a better job without you.

True AI systems are going to be much better at investigation and diagnosis and therapy than any human, especially because of the absence of confirmation bias and beliefs. This programmed proxy is absolutely not the way to proceed.

There's an arrogance in this approach of programming a robot to carry out the instructions of the psychologists, a presumption that all technology is good for is carrying out our orders to the letter. AI systems have already gone way beyond that. Google's Alpha neural network is currently exploring protein folding on a level humans can barely understand. A few months ago they gave it the day off and invited it to teach itself chess. After a few hours of playing against itself it became stronger than any programmed chess engine (and those were already way stronger than any human player). All Google's programmers did was tell it the rules of the game. It figured out how to play the game all on its own, just by doing it. Just imagine what that network might achieve if unleashed on ME bio data banks.

The sort of approach described in this article and paper will seem comical to people even 20 years from now.

 

[Trish]

No way that robots can replace humans. Unless maybe the robots wear blue cardigans or some similarly impressive equivalent.

On the plus side, all the suddenly unemployed therapists will have "someone" to talk to.

The sort of approach described in this article and paper will seem comical to people even 20 years from now.

No need to wait 20 years. It's comical now.

Thanks all, I love this forum!

 

[Forbin]

I swear I hear my robot therapist coming up the steps, but he never seems to get here.

 

[Trish]

That is so funny, @Forbin. The funniest bit was after the robot fell and they put a screen around it - as if it were a human who had to be hidden to give it privacy.

 

[Barry]

"The next stage is to undertake a quantitative study, where we can measure whether participants felt that the intervention actually increased their activity levels."

Don't you just love the way these folk so easily slip into their favourite fantasy - that they can measure people's feelings with sufficient objectivity to produce validly quantitative readings.